Clarify factual positioning and performance claims in README and related work

[vjovanov]

Discussion status

Scoped documentation proposal. Keep human-discussion until the owner resolves the open editorial choices and releases it for implementation. This ticket has no release or 1.0 dependency.

Outcome

A reader can tell what Grund checks and retrieves, how it fits beside link checkers and requirements-traceability tools, and what its performance numbers actually measure. The README links to a factual, maintainable explanation without promising unimplemented capabilities or presenting unlike workloads as a general speed comparison.

This consolidates the remaining work in RM-positioning and RM-positioning-trace-tools. They share the same reader, documentation surfaces, and evidence review; separate tickets would duplicate that work.

Already shipped; exclude from new work

At main df506acfae6d39169e7e2cdabd2327f669a21c3d:

• README line 20 already contains the requested “Lychee is the link checker; grund is the intent checker” paragraph and says both belong in CI. Do not reimplement that milestone by adding another slogan block. Accuracy corrections to existing claims remain in scope.
• The README already demonstrates section-level citations and the resolver's lead, brief, table-of-contents, and full-body modes. Reuse that explanation instead of repeating it.
• docs/related-work/REL-traceability-tools.md already contains the comparison and primary-project links, delivered with Add a REL kind for related work: the ideas grund descends from and the tools beside it #384. The remaining work is to review and connect it, not create a parallel comparison document.
• The instruction-count harness and recorded benchmark material already exist. This ticket adds no benchmark machinery, thresholds, measurements, or engine features.

Remaining scope

1. Explain the performance evidence honestly. Add a short explanation beside the README throughput claim or its immediate supporting text, linking the local wall-clock report and instruction-count methodology. Distinguish the dated local throughput snapshot, the historical committed instruction-count snapshot, and current generated-fixture/base-branch CI comparisons. Resolve the report's contradictory “release-blocking meter” wording: the current architecture explicitly says the regression limits are not wired to fail the build. Describe instruction count as a repeatable workload-cost proxy with its binary/input/build assumptions; do not turn it into a universal latency guarantee or claim an active regression gate that is absent. Do not refresh numbers merely to complete this documentation task.
2. Explain the traceability relationship through concrete reader tasks. Add a concise README entry pointing to the existing related-work document. Ground Grund's own claims in shipped section-addressed retrieval, citation validation, and supported checked constraints. Avoid generic assertions that every neighbouring tool serves only coverage reports or requires every clause to be a separate item. Verify any retained competitor claim against current primary documentation, recording the relevant source/version/date, or remove/qualify it when evidence is insufficient.
3. Audit the existing comparison rather than require a marketing matrix. Prefer a short task-oriented explanation. Keep a table only where supported, consistently defined dimensions help a reader choose a tool; exact Grund flag spelling is not a neutral test of another tool's retrieval capability. Do not require six rows, creation years, checkmark rankings, or a copied table in the README. The related-work document is the detailed home.
4. Align boundary claims with the actual contract. The roadmap's proposed blanket ban on schema-level custom checks is inconsistent with existing grounded chapter/citation rules. Explain the supported declarative constraints and the limits on arbitrary executable checks and severity customization precisely. Review references to non-goals; correct unsupported inferences without creating new product policy. v2 schema: enforce ordered fields, scalar values, closed shapes, and declaration forms #458 and After 1.0: express place citation and grounding constraints as grounded rule declarations #468 are open design work, not shipped capabilities and not prerequisites for this docs cleanup. If mentioned, distinguish their proposed future scope explicitly.
5. Remove unsupported future parity promises from the touched positioning material. Do not retain the related-work table's promised gap-report capability or the roadmap-derived “coverage parity is one shipping milestone away” line. Any remaining coverage description must name the exact current query/report behavior and its limits. This does not authorize a gap-report feature, its implementation, or a replacement proposal.

Apply equivalent factual corrections to an existing landing page only if one is identified; creating a site is out of scope.

Acceptance criteria

• README readers can follow one short path to the detailed comparison and benchmark evidence, without duplicate introductory copy.
• Every new or retained competitor capability claim in the touched comparison has current primary-source support and a clear definition; uncertain assertions are qualified or removed.
• Performance copy distinguishes measured workload, date/build assumptions, proxy versus elapsed time, and reporting versus enforced thresholds. It does not imply fair interchangeable workloads merely from a timing ratio.
• The related-work document no longer promises gap/coverage parity or contradicts shipped checked-rule support. Proposed v2 work is never presented as already available.
• Documentation links/citations point to the most specific supporting source and the documents agree with one another. No CLI, core, configuration, benchmark code, or release change is part of completion.

Sources

• Roadmap: positioning and traceability positioning are historical inputs to be corrected against shipped facts, not copy to implement literally.
• README at the reviewed commit.
• Existing related work.
• Local benchmark report, benchmark architecture, and CI regression-limit status.
• Checked rules, non-goals, and README requirements.

Open editorial decisions

• Is concise prose plus the existing related-work link enough, or does a smaller evidence-backed table serve an identifiable reader decision?
• Should the dated throughput badge remain with explicit context or be retired? No new benchmark run is required by this ticket.
• Which unsupported historical comparison claims should be removed entirely rather than maintained with recurring verification?

No required competitor slogan, new non-goal, release date, or coverage-parity commitment is pre-approved here.

[vjovanov]

Claimed by vjovanov@kung-fu-workstation on 2026-10-05 22:51Z · released on 2026-10-06 01:10Z (completed).

[vjovanov]

Record · 41 transitions · now completed · last 2026-10-06 00:44Z
Holder vjovanov@kung-fu-workstation · since 2026-10-05 22:51Z

1 · supervising → supervising 2026-10-05 22:53Z · `PATH: planned and PLANNER: smart remain as recorded in decision.md; do not rewrite that decision. The planner's existing Target cdx[high]:openai:gpt-6.1-sol remains appropriate for reconciling grounded documentation across four surfaces under the quoted OpenAI-only ruling. No escalation or metadata edit is needed. Clarify is completed and Agora cancelled; the planner has no outstanding prerequisite. Release exactly ticket.plan through its task-wide brief; dependencies, specify, implement, reviews, fixes, green, describe, and ship remain unbriefed.`

plan runtime/supervision/plan.md (not shown; this comment is at its size limit)

decision runtime/supervision/decision.md (not shown; this comment is at its size limit)

2 · ticket.triage · intake → classifying 2026-10-05 22:53Z

ticket runtime/ticket/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.json (not shown; this comment is at its size limit)

3 · ticket.triage · classifying → labelling 2026-10-05 22:53Z · `KIND: usability`

kind runtime/ticket/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.kind.md (not shown; this comment is at its size limit)

4 · ticket.triage · labelling → triage 2026-10-05 22:53Z

5 · ticket.triage · triage → assessing 2026-10-05 22:53Z

candidates runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.candidates.md

Candidates for github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage

Tool: grund · Repo: agent-grounds/grund · Kind: usability · Query: clarify OR factual OR positioning OR performance OR claims OR readme

Upstream issues

A broad OR search - expect unrelated hits. Judge each on the
behaviour it describes, not on shared words.

• [CLOSED] Add a REL kind for related work: the ideas grund descends from and the tools beside it #384 Add a REL kind for related work: the ideas grund descends from and the tools beside it
  Add a REL kind for related work: the ideas grund descends from and the tools beside it #384
• [CLOSED] A reproducible factual claim should be a checked edge, not prose beside a citation that resolves #229 A reproducible factual claim should be a checked edge, not prose beside a citation that resolves
  A reproducible factual claim should be a checked edge, not prose beside a citation that resolves #229
• [OPEN] Towards grund 1.0: the v2 configuration format, the concern split, and the releases between here and 1.0 #347 Towards grund 1.0: the v2 configuration format, the concern split, and the releases between here and 1.0
  Towards grund 1.0: the v2 configuration format, the concern split, and the releases between here and 1.0 #347
• [OPEN] CLI: move report preferences to run flags and implement v2 formatting presentation #459 CLI: move report preferences to run flags and implement v2 formatting presentation
  CLI: move report preferences to run flags and implement v2 formatting presentation #459
• [CLOSED] Retire "ref" from functional-spec prose in favour of "citation" #309 Retire "ref" from functional-spec prose in favour of "citation"
  Retire "ref" from functional-spec prose in favour of "citation" #309
• [CLOSED] missing-index-entry prints the same "is not listed" sentence whether the index has no row for the ID or a markerless one the reader can see #297 missing-index-entry prints the same "is not listed" sentence whether the index has no row for the ID or a markerless one the reader can see
  missing-index-entry prints the same "is not listed" sentence whether the index has no row for the ID or a markerless one the reader can see #297
• [CLOSED] Sweep the retired words out of the functional-spec prose, one pull request per word #291 Sweep the retired words out of the functional-spec prose, one pull request per word
  Sweep the retired words out of the functional-spec prose, one pull request per word #291
• [CLOSED] grund init --docs --force leaves numbered ID examples under slug-only format #266 grund init --docs --force leaves numbered ID examples under slug-only format
  grund init --docs --force leaves numbered ID examples under slug-only format #266
• [OPEN] Agent guidance: render each kind’s shape and rules at its common ancestor directory #460 Agent guidance: render each kind’s shape and rules at its common ancestor directory
  Agent guidance: render each kind’s shape and rules at its common ancestor directory #460

(the sibling-ticket index that follows is the plan itself; omitted)

6 · ticket.triage · assessing → route-assessment 2026-10-05 22:57Z · `VERDICT: none`

verdict runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.verdict.md

No candidate describes the remaining documentation corrections in #475. The originating OPEN issue #475 is this ticket's source, not another candidate against which to deduplicate it. Issue bodies and current states were fetched read-only on 2026-10-06 and saved in github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.upstream.json beside this file.

The closest candidate is CLOSED #384, #384. It added the REL kind, related-work home, grounding directions, and seed comparison. Its traceability paragraph identifies a missing home for the comparison, rather than the release-blocking benchmark wording, unsupported capability rankings, blanket custom-check exclusion, or future coverage-parity promise. That home exists in this checkout. #475 expressly excludes recreating it. These corrections are not a recurrence of #384's missing-home behavior.

CLOSED #229, #229, proposes making reproducible factual claims checked graph edges. It uses stale measurements and transcripts as examples, but asks for checker machinery. #475 instead corrects specific positioning prose and expressly excludes machinery and new capabilities. A general mechanism for preventing stale prose does not track this particular correction.

OPEN #347, #347, is the v2/1.0 design and release umbrella. #475 has no release dependency and must describe the shipped contract. OPEN #459, #459, changes CLI report flags and v2 formatting presentation. Neither asks to correct these comparison or performance statements.

CLOSED #309, #309, and CLOSED #291, #291, sweep retired terminology from functional-spec prose. This ticket concerns accuracy of capabilities and measurements, not substituting settled vocabulary.

CLOSED #297, #297, concerns indistinguishable missing-index diagnostics for absent versus markerless entries. CLOSED #266, #266, concerns numbered ID examples generated by init under a slug-only configuration. Their concrete triggers and faulty outputs are unrelated to the passages reported here.

OPEN #460, #460, concerns placement and rendering of generated agent guidance at kind-directory common ancestors. It does not concern reader positioning or benchmark evidence.

The candidates file reports no sibling-ticket index. Directly reading this execution's index.rhei.md and tasks/01-ticket.md finds one originating ticket and its lifecycle steps, not a second ticket describing the behavior. The execution roots under this checkout's panta contain only this rhei. No sibling match was found in that available scope.

… cut at 3000 characters; the whole file is runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.verdict.md

7 · ticket.triage · route-assessment → validating 2026-10-05 22:57Z

8 · ticket.triage · validating → reproducing 2026-10-05 22:57Z

9 · ticket.triage · reproducing → run-reproducer 2026-10-05 23:01Z · `REPRO: yes`

repro runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.repro.md

Validated grund issue #475 at checkout df506ac on 2026-10-06 (Europe/Zurich). The reported documentation defects remain. This reproduces misleading published documentation, not a CLI malfunction. No benchmark measurements, specification changes, implementation edits, or upstream writes were made.

KIND: usability
VERDICT: none

The prior verdict was read first: no supplied candidate or available sibling ticket matches this correction. #384 created the related-work home that already exists; #229 proposes checker machinery outside this ticket's scope. The current issue is the source, not a duplicate candidate.

Run from the checkout root, with Bash and grep installed; no network or build is needed:

panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh

Observed exit status: 1. The script prints each expectation and the exact offending line before reporting 5 reported documentation defects found. Full output and exit statuses are in runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/transcript.txt, relative to this execution root.

• docs/benchmarks.md:7 says “The release-blocking meter is Callgrind instruction count” and, in the same paragraph, says the limits are not wired up. Expected: describe comparison/reporting without claiming an active instruction-regression gate. §AR-ci.5.2 docs/architecture/AR-ci.md:113 explicitly says the intended limits do not fail the build. .github/workflows/ci.yml:194–207 supplies a base-branch comparison without limit flags; crates/grund-cli/benches/instructions.rs contains no RegressionConfig.
• docs/related-work/REL-traceability-tools.md:28 gives Grund's coverage-report cell an hourglass and a gap-report roadmap pointer. Expected: describe exact shipped queries and constraints, without promising that future feature. §FS-cover.2 docs/functional-spec/FS-cover.md:27 groups citations by scanned file, including files with no citations, and does not judge sufficient coverage or co-change. Existing per-target and inbound checked constraints also matter: §FS-rules.3.3 docs/functional-spec/FS-rules.md:172 and §FS-rules.3.4 docs/functional-spec/FS-rules.md:188.
• docs/roadmap.md:210 says “Coverage parity is one shipping milestone away.” Expected: remove this unapproved parity promise from positioning; an unspecified future report cannot establish parity with other tools.
  … cut at 3000 characters; the whole file is runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.repro.md

run.sh runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh

#!/usr/bin/env bash
set -uo pipefail

Read-only detector for the specific documentation defects in grund issue #475.

Run from the checkout root. No benchmarks, network, fixtures, or writes needed.

Ground and limitations are recorded in the validation export and repro.md.

if (( $# != 0 )); then
echo 'Cannot run: this reproducer accepts no arguments or flags.' >&2
exit 2
fi
if ! command -v grep >/dev/null 2>&1; then
echo 'Cannot run: grep is required.' >&2
exit 2
fi
for file in README.md docs/benchmarks.md docs/roadmap.md
docs/related-work/REL-traceability-tools.md; do
if [[ ! -f "$file" || ! -r "$file" ]]; then
echo "Cannot run: missing or unreadable $file; run from the grund checkout root." >&2
exit 2
fi
done

failures=0
check_absent() {
local file="$1" pattern="$2" expected="$3" observed status
observed=$(grep -nE -- "$pattern" "$file")
status=$?
if (( status > 1 )); then
echo "Cannot run: grep failed while reading $file (exit $status)." >&2
exit 2
fi
echo "Expected: $expected"
if (( status == 0 )); then
echo "Observed FAIL $file:$observed"
failures=$((failures + 1))
else
echo "Observed PASS $file: reported misleading passage is absent."
fi
}

check_absent docs/benchmarks.md
'The release-blocking meter is Callgrind instruction count'
'Instruction-count comparison is described as reporting while regression limits are unwired.'
check_absent docs/related-work/REL-traceability-tools.md
'^| **grund** |.*⏳.*RM-gap-report'
'The comparison describes shipped capability without promising a future gap report.'
check_absent docs/roadmap.md
'Coverage parity is one shipping milestone away'
'Positioning makes no future coverage-parity promise.'
check_absent docs/roadmap.md
'schema-level custom check rules (would require severity / exit-code config'
'Positioning distinguishes supported declarative constraints from severity customization.'

echo 'Expected: README links to the existing detailed traceability comparison.'
grep -qE -- '(](|]:[[:space:]]*)[^[:space:]]*docs/related-work/REL-traceability-tools.md([#)[:space:]]|$)' README.md
status=$?
if (( status > 1 )); then
echo "Cannot run: grep failed while reading README.md (exit $status)." >&2
exit 2
elif (( status == 1 )); then
echo 'Observed FAIL README.md: no Markdown link to docs/related-work/REL-traceability-tools.md.'
failures=$((failures + 1))
else
echo 'Observed PASS README.md: comparison link found.'
fi

echo "$failures reported documentation defects found."
if (( failures != 0 )); then
exit 1
fi
exit 0

10 · ticket.triage · run-reproducer → validated 2026-10-05 23:01Z

run.log runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.log

Reproducer for github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage

script: /home/vjovanov/ag/grund/fix/issue-475/panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh
checkout: /home/vjovanov/ag/grund/fix/issue-475
timeout: 20m
started: 2026-10-05T23:01:15Z

Expected: Instruction-count comparison is described as reporting while regression limits are unwired.
Observed FAIL docs/benchmarks.md:7:The release-blocking meter is Callgrind instruction count, not wall-clock time (§DA-benchmark-instruction-counting). Pull-request CI compares against the current base branch and records the Ir comparison in its summaries; the limits are not wired up, so growth does not yet fail the build (§AR-ci.5.2); the table below is the committed human-readable snapshot from cargo bench -p grund --features bench --locked --bench instructions -- --save-summary=json on 2026-05-20.
Expected: The comparison describes shipped capability without promising a future gap report.
Observed FAIL docs/related-work/REL-traceability-tools.md:28:| grund | 2026 | ✅ | ✅ | ✅ | ✅ | ⏳ §RM-gap-report | ✅ |
Expected: Positioning makes no future coverage-parity promise.
Observed FAIL docs/roadmap.md:210:- Coverage parity is one shipping milestone away. §RM-gap-report inverts the §FS-cover index and answers "which IDs are uncovered?" — the column that today reads ⏳ in that matrix.
Expected: Positioning distinguishes supported declarative constraints from severity customization.
Observed FAIL docs/roadmap.md:212:A short "we deliberately don't" footnote points at §FS-non-goals and names three features the neighbours have that grund will not grow: ReqIF / OFT interchange (would import a foreign citation grammar and break the "two installs agree" contract, §FS-non-goals.13), schema-level custom check rules (would require severity / exit-code config, §FS-non-goals.9), and HTML/PDF publishing (a third first-party surface, §FS-non-goals.12).
Expected: README links to the existing detailed traceability comparison.
… cut at 3000 characters; the whole file is runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.log

11 · ticket.triage · validated → completed 2026-10-05 23:01Z

validation runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/validation.md

Validation of github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage: Clarify factual positioning and performance claims in README and related work

• Kind: usability
• Origin: issue agent-grounds/grund#475 (Clarify factual positioning and performance claims in README and related work #475)
• Verdict: none
• Reproduced: yes
• Size: not stated

The reproduction

Validated grund issue #475 at checkout df506ac on 2026-10-06 (Europe/Zurich). The reported documentation defects remain. This reproduces misleading published documentation, not a CLI malfunction. No benchmark measurements, specification changes, implementation edits, or upstream writes were made.

KIND: usability
VERDICT: none

The prior verdict was read first: no supplied candidate or available sibling ticket matches this correction. #384 created the related-work home that already exists; #229 proposes checker machinery outside this ticket's scope. The current issue is the source, not a duplicate candidate.

Run from the checkout root, with Bash and grep installed; no network or build is needed:

panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh

Observed exit status: 1. The script prints each expectation and the exact offending line before reporting 5 reported documentation defects found. Full output and exit statuses are in runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/transcript.txt, relative to this execution root.

• docs/benchmarks.md:7 says “The release-blocking meter is Callgrind instruction count” and, in the same paragraph, says the limits are not wired up. Expected: describe comparison/reporting without claiming an active instruction-regression gate. §AR-ci.5.2 docs/architecture/AR-ci.md:113 explicitly says the intended limits do not fail the build. .github/workflows/ci.yml:194–207 supplies a base-branch comparison without limit flags; crates/grund-cli/benches/instructions.rs contains no RegressionConfig.
• docs/related-work/REL-traceability-tools.md:28 gives Grund's coverage-report cell an hourglass and a gap-report roadmap pointer. Expected: describe exact shipped queries and constraints, without promising that future feature. §FS-cover.2 docs/functional-spec/FS-cover.md:27 groups citations by scanned file, including files with no citations, and does not judge sufficient coverage or co-change. Existing per-target and inbound checked constraints also matter: §FS-rules.3.3 docs/functional-spec/FS-rules.md:172 and §FS-rules.3.4 docs/functional-spec/FS-rules.md:188.
• docs/roadmap.md:210 says “Coverage parity is one shipping milestone away.” Expected: remove this unapproved parity promise from positioning; an unspecified future report cannot establish parity with other tools.
  … cut at 3000 characters; the whole file is runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/validation.md

12 · ticket.agora · agora → cancelled

13 · supervising → supervising 2026-10-05 22:53Z

14 · ticket.clarify · clarifying → asking 2026-10-05 23:06Z · `QUESTIONS: ready`

questions runtime/clarify/questions-1.md

The README currently shows grund check: ~722k LoC/s without a date beside it. Its linked report describes a local run from 2026-05-20, while also calling instruction counts “release-blocking” even though their regression limits do not fail CI. After this correction, readers would reach that dated evidence and the current reporting methodology from a short README explanation, and reach the existing traceability comparison from one link. The factual corrections are clear; your ticket reserves the following publication choices.

1. Comparison format. Should the existing related-work document explain concrete tasks in concise prose, or retain a smaller table with consistently defined, primary-source-backed dimensions? For example, reply “prose” to replace the rankings with an explanation of reading a cited section and checking citations and declarative constraints; reply “table” to request a compact comparison organized around a reader decision, such as retrieving a requirement by ID. Either choice gets one short README link and reuses the existing introduction and retrieval examples.

   Why this needs an answer: The ticket expressly reserves the format, so the proposal cannot treat either format as approved.

   Assumption: Use task-oriented prose in the existing document, without a ranking matrix.

2. Throughput badge. Should the undated ~722k LoC/s badge be retired, or kept with nearby date, workload, build and measurement context? Reply “retire” to remove the badge while preserving the dated report and a short README performance-evidence link; reply “keep with context” to retain the number and make its local snapshot assumptions visible beside it. Neither answer triggers new measurements. Both distinguish local elapsed time, the historical instruction-count snapshot, and current generated-fixture/base-branch reporting.

   Why this needs an answer: Keeping a headline performance number versus directing readers to historical evidence changes the publication scope you reserved.

   Assumption: Retire the badge, preserve the report, and link and explain the evidence briefly.

3. Historical competitor assertions. May unsupported blanket claims, creation dates, installation rankings and exact-Grund-flag capability tests be removed, keeping only reader-useful claims with current primary-source/version/date evidence, or do you want particular assertions maintained and verified? For example, reply “remove unsupported claims” to drop the blanket “They model each clause as its own atomic item”; reply “maintain the installation comparison” to request a defined, sourced comparison of installation requirements, without preserving an unsupported ranking.

   Why this needs an answer: The proposal needs to know which additional comparison claims you want maintained rather than removed.

… cut at 3000 characters; the whole file is runtime/clarify/questions-1.md

questions runtime/clarify/questions.md

The README currently shows grund check: ~722k LoC/s without a date beside it. Its linked report describes a local run from 2026-05-20, while also calling instruction counts “release-blocking” even though their regression limits do not fail CI. After this correction, readers would reach that dated evidence and the current reporting methodology from a short README explanation, and reach the existing traceability comparison from one link. The factual corrections are clear; your ticket reserves the following publication choices.

1. Comparison format. Should the existing related-work document explain concrete tasks in concise prose, or retain a smaller table with consistently defined, primary-source-backed dimensions? For example, reply “prose” to replace the rankings with an explanation of reading a cited section and checking citations and declarative constraints; reply “table” to request a compact comparison organized around a reader decision, such as retrieving a requirement by ID. Either choice gets one short README link and reuses the existing introduction and retrieval examples.

   Why this needs an answer: The ticket expressly reserves the format, so the proposal cannot treat either format as approved.

   Assumption: Use task-oriented prose in the existing document, without a ranking matrix.

2. Throughput badge. Should the undated ~722k LoC/s badge be retired, or kept with nearby date, workload, build and measurement context? Reply “retire” to remove the badge while preserving the dated report and a short README performance-evidence link; reply “keep with context” to retain the number and make its local snapshot assumptions visible beside it. Neither answer triggers new measurements. Both distinguish local elapsed time, the historical instruction-count snapshot, and current generated-fixture/base-branch reporting.

   Why this needs an answer: Keeping a headline performance number versus directing readers to historical evidence changes the publication scope you reserved.

   Assumption: Retire the badge, preserve the report, and link and explain the evidence briefly.

3. Historical competitor assertions. May unsupported blanket claims, creation dates, installation rankings and exact-Grund-flag capability tests be removed, keeping only reader-useful claims with current primary-source/version/date evidence, or do you want particular assertions maintained and verified? For example, reply “remove unsupported claims” to drop the blanket “They model each clause as its own atomic item”; reply “maintain the installation comparison” to request a defined, sourced comparison of installation requirements, without preserving an unsupported ranking.

   Why this needs an answer: The proposal needs to know which additional comparison claims you want maintained rather than removed.

… cut at 3000 characters; the whole file is runtime/clarify/questions.md

15 · ticket.clarify · asking → completed 2026-10-05 23:06Z · `CLARIFICATION: answered`

answers runtime/clarify/answers.md

Questions and answers on #475

Round 1 of 2

Asked at: #475 (comment)

The README currently shows grund check: ~722k LoC/s without a date beside it. Its linked report describes a local run from 2026-05-20, while also calling instruction counts “release-blocking” even though their regression limits do not fail CI. After this correction, readers would reach that dated evidence and the current reporting methodology from a short README explanation, and reach the existing traceability comparison from one link. The factual corrections are clear; your ticket reserves the following publication choices.

1. Comparison format. Should the existing related-work document explain concrete tasks in concise prose, or retain a smaller table with consistently defined, primary-source-backed dimensions? For example, reply “prose” to replace the rankings with an explanation of reading a cited section and checking citations and declarative constraints; reply “table” to request a compact comparison organized around a reader decision, such as retrieving a requirement by ID. Either choice gets one short README link and reuses the existing introduction and retrieval examples.

   Why this needs an answer: The ticket expressly reserves the format, so the proposal cannot treat either format as approved.

   Assumption: Use task-oriented prose in the existing document, without a ranking matrix.

2. Throughput badge. Should the undated ~722k LoC/s badge be retired, or kept with nearby date, workload, build and measurement context? Reply “retire” to remove the badge while preserving the dated report and a short README performance-evidence link; reply “keep with context” to retain the number and make its local snapshot assumptions visible beside it. Neither answer triggers new measurements. Both distinguish local elapsed time, the historical instruction-count snapshot, and current generated-fixture/base-branch reporting.

   Why this needs an answer: Keeping a headline performance number versus directing readers to historical evidence changes the publication scope you reserved.

   Assumption: Retire the badge, preserve the report, and link and explain the evidence briefly.

3. Historical competitor assertions. May unsupported blanket claims, creation dates, installation rankings and exact-Grund-flag capability tests be removed, keeping only reader-useful claims with current primary-source/version/date evidence, or do you want particular assertions maintained and verified? For example, reply “remove unsupported claims” to drop the blanket “They model each clause as its own atomic item”; reply “maintain the installation comparison” to request a defined, sourced comparison of installation requirements, without preserving an unsupported ranking.

… cut at 3000 characters; the whole file is runtime/clarify/answers.md

clarification runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.clarify/clarification.md

Clarification of #475

CLARIFICATION: answered

Questions and answers on #475

Round 1 of 2

Asked at: #475 (comment)

The README currently shows grund check: ~722k LoC/s without a date beside it. Its linked report describes a local run from 2026-05-20, while also calling instruction counts “release-blocking” even though their regression limits do not fail CI. After this correction, readers would reach that dated evidence and the current reporting methodology from a short README explanation, and reach the existing traceability comparison from one link. The factual corrections are clear; your ticket reserves the following publication choices.

1. Comparison format. Should the existing related-work document explain concrete tasks in concise prose, or retain a smaller table with consistently defined, primary-source-backed dimensions? For example, reply “prose” to replace the rankings with an explanation of reading a cited section and checking citations and declarative constraints; reply “table” to request a compact comparison organized around a reader decision, such as retrieving a requirement by ID. Either choice gets one short README link and reuses the existing introduction and retrieval examples.

   Why this needs an answer: The ticket expressly reserves the format, so the proposal cannot treat either format as approved.

   Assumption: Use task-oriented prose in the existing document, without a ranking matrix.

2. Throughput badge. Should the undated ~722k LoC/s badge be retired, or kept with nearby date, workload, build and measurement context? Reply “retire” to remove the badge while preserving the dated report and a short README performance-evidence link; reply “keep with context” to retain the number and make its local snapshot assumptions visible beside it. Neither answer triggers new measurements. Both distinguish local elapsed time, the historical instruction-count snapshot, and current generated-fixture/base-branch reporting.

   Why this needs an answer: Keeping a headline performance number versus directing readers to historical evidence changes the publication scope you reserved.

   Assumption: Retire the badge, preserve the report, and link and explain the evidence briefly.

3. Historical competitor assertions. May unsupported blanket claims, creation dates, installation rankings and exact-Grund-flag capability tests be removed, keeping only reader-useful claims with current primary-source/version/date evidence, or do you want particular assertions maintained and verified? For example, reply “remove unsupported claims” to drop the blanket “They model each clause as its own atomic item”; reply “maintain the installation comparison” to request a defined, sourced comparison of installation requirements, without preserving an unsupported ranking.

… cut at 3000 characters; the whole file is runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.clarify/clarification.md

16 · supervising → supervising 2026-10-05 22:53Z

17 · ticket.plan · planning → discussing 2026-10-05 23:36Z

proposal runtime/plan/proposal-1.md

1. What changes for the reader

Open README.md and follow the grund check: ~722k LoC/s badge. Today it leads to a dated report with contradictory “release-blocking” wording; the README also gives no link to the existing traceability comparison. After this correction, the badge is retired and two short entries lead to the evidence and comparison. Sample performance wording:

The benchmark report preserves a 2026-05-20 local timing run and a historical instruction-count snapshot. Current CI compares instruction counts on generated fixtures against the PR base branch. Counts proxy workload cost, not elapsed time; regression limits are not enforced.

Readers can distinguish what Grund retrieves and checks from what other tools offer, and historical measurements from current reporting.

2. Design and reuse

Your 👍 accepted “Use task-oriented prose in the existing document, without a ranking matrix,” “Retire the badge, preserve the report, and link and explain the evidence briefly,” and “Remove the unsupported historical assertions and retain only narrow, useful comparisons that can be supported by current primary sources.”

The existing related-work page carries the detailed explanation; the README reuses its introduction and retrieval examples. Its existing link-checker paragraph will say Lychee checks Markdown and HTML links, while Grund resolves marked citations and checks declared constraints. The performance explanation follows generated-workload methodology (§AR-benchmarks.1.1), instruction-count rationale (§AR-benchmarks.2), and reporting status (§AR-ci.5.2).

The retained competitor claims have this finite source ledger, checked 2026-10-06:

• Lychee checks Markdown and HTML: official documentation, unversioned.
• OpenFastTrace traces specification items through source tags and produces HTML reports: user guide, report instructions; unversioned guides, current release 4.10.0.
  … cut at 3000 characters; the whole file is runtime/plan/proposal-1.md

proposal runtime/plan/proposal.md

1. What changes for the reader

Open README.md and follow the grund check: ~722k LoC/s badge. Today it leads to a dated report with contradictory “release-blocking” wording; the README also gives no link to the existing traceability comparison. After this correction, the badge is retired and two short entries lead to the evidence and comparison. Sample performance wording:

The benchmark report preserves a 2026-05-20 local timing run and a historical instruction-count snapshot. Current CI compares instruction counts on generated fixtures against the PR base branch. Counts proxy workload cost, not elapsed time; regression limits are not enforced.

Readers can distinguish what Grund retrieves and checks from what other tools offer, and historical measurements from current reporting.

2. Design and reuse

Your 👍 accepted “Use task-oriented prose in the existing document, without a ranking matrix,” “Retire the badge, preserve the report, and link and explain the evidence briefly,” and “Remove the unsupported historical assertions and retain only narrow, useful comparisons that can be supported by current primary sources.”

The existing related-work page carries the detailed explanation; the README reuses its introduction and retrieval examples. Its existing link-checker paragraph will say Lychee checks Markdown and HTML links, while Grund resolves marked citations and checks declared constraints. The performance explanation follows generated-workload methodology (§AR-benchmarks.1.1), instruction-count rationale (§AR-benchmarks.2), and reporting status (§AR-ci.5.2).

The retained competitor claims have this finite source ledger, checked 2026-10-06:

• Lychee checks Markdown and HTML: official documentation, unversioned.
• OpenFastTrace traces specification items through source tags and produces HTML reports: user guide, report instructions; unversioned guides, current release 4.10.0.
  … cut at 3000 characters; the whole file is runtime/plan/proposal.md

18 · ticket.plan · discussing → completed 2026-10-05 23:36Z

response runtime/plan/response.md

The author approved the plan by reaction

Round 1 of 3, posted at: #475 (comment)

vjovanov reacted 👍 to round 1 and wrote nothing further.
That is approval of the plan as proposed: implement it as written.
No part of it was amended, and nothing was added - where the plan left
something to your judgement, it is still yours.

19 · supervising → supervising 2026-10-05 22:53Z

20 · ticket.dependencies · dependencies → completed 2026-10-05 23:57Z · `DEPENDENCIES: clear`

dependencies runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.dependencies/dependencies.md

Dependencies for #475

No prerequisite ticket was declared by the approved proposal.

DEPENDENCIES: clear

21 · supervising → supervising 2026-10-05 22:53Z

22 · ticket.specify · specify → opening 2026-10-06 00:11Z · `PR: https://github.com//pull/476`

contract runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.specify/contract.md

Specification contract for grund #475

The README and supporting documents must provide sourced, task-oriented positioning and distinguish archival measurements from current reporting. This state writes the requirement and tests only. Push and draft-PR creation belong to the following opening program; no PR number or URL exists at this handoff. Later steps adopt runtime/spec/pr.md.

Authority and scope

The approved contract is runtime/plan/proposal.md, with the posted copy in runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.plan/proposal.md. The owner's approval at #475 (comment) is recorded in runtime/plan/response.md. The dependency export declares an empty prerequisite set and DEPENDENCIES: clear.

Only these repository files change:

• docs/requirements/REQ-readme.md: extended §REQ-readme.1 docs/requirements/REQ-readme.md:5; new §REQ-readme.evidence docs/requirements/REQ-readme.md:21.
• tests/integration/test_benchmark_report_claims.py.
• tests/integration/test_related_work.py.

Existing numbered requirement addresses remain intact. No shipped functional specification, architecture policy, non-goal, verdict, CLI/core/configuration, benchmark generator, measurement, changelog, site or release change is made.

Commit: 5d0bf1c70b1b41804d18f74d91a922cb974c1a33 — Require factual positioning and performance evidence, pinned by tests.
It contains only the requirement and the two test files, on the existing fix/issue-475 branch.

Reproduction and command

First ran the original detector, before writing tests:

bash panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh

Exit 1, quoting its summary: 5 reported documentation defects found. It printed the report's release-blocking claim, the comparison's future gap-report cell, the roadmap's coverage-parity promise and schema-check prohibition, and the missing README comparison link. The validation export and adjacent triage transcript carry the original exact passages.

One command runs the complete targeted contract from the checkout root:

PYTHONPATH=tests/integration python -m unittest test_benchmark_report_claims test_related_work -v

Exit 1. Exact final summary:

Ran 27 tests in 0.097s

FAILED (failures=26)

Ten repository-document test methods fail, producing 26 failure events including subtests; seventeen methods pass. Full final output is runtime/spec/targeted-tests.log. Earlier discovery runs are runtime/spec/benchmark-tests.log and runtime/spec/related-work-tests.log. No expected-failure decorators, skipped assertions or gate bypasses hide the red contract.

New benchmark tests and current failures

File: tests/integration/test_benchmark_report_claims.py. Class: TheBenchmarkComparisonIsRecordedNotGating.

… cut at 3000 characters; the whole file is runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.specify/contract.md

pr runtime/spec/pr.md

PR: #476

23 · ticket.specify · opening → completed 2026-10-06 00:11Z

24 · supervising → supervising 2026-10-05 22:53Z

25 · ticket.implement · implement → completed 2026-10-06 00:20Z

report runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.implement/report.md

{
  "summary": "README readers now reach the existing task-oriented traceability comparison and archival/current performance evidence without an undated throughput badge. The comparison documents shipped section retrieval, citation validation, per-file citation grouping and declarative constraints with precise limits and a dated primary-source ledger. The benchmark report distinguishes historical wall-clock and instruction-count snapshots from current generated-fixture/base-branch reporting, preserves measurements and provenance, explains binary/build/PGO assumptions and unlike workloads, and directs future output outside the archive. The roadmap agrees without rankings, coverage-parity promises or unsupported policy inferences.",
  "root_cause": "As the approved plan diagnosed, unsupported comparison generalisations and future promises were mixed with shipped facts, while archival performance evidence was presented as current positioning and an active regression gate.",
  "commits": [
    "38cceb5c95b67bd0c3c42fcb7f85bfd41d0c978d"
  ],
  "files": [
    "README.md",
    "docs/related-work/REL-traceability-tools.md",
    "docs/benchmarks.md",
    "docs/roadmap.md",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/implement/source-audit.md",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/implement/lychee.txt",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/implement/oft-guide.txt",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/implement/oft-reports.txt",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/implement/oft-release.txt",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.implement/report.md",
    "panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/results/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.implement.md"
  ],
  "tests": [
    "test_benchmark_report_claims.TheBenchmarkComparisonIsRecordedNotGating.test_public_copy_has_no_active_instruction_regression_gate",
    "test_benchmark_report_claims.TheBenchmarkComparisonIsRecordedNotGating.test_readme_retires_the_throughput_badge",
    "test_benchmark_report_claims.TheBenchmarkComparisonIsRecordedNotGating.test_readme_links_to_performance_evidence_and_current_reporting",
    "test_benchmark_report_claims.TheBenchmarkComparisonIsRecordedNotGating.test_report_separates_historical_measurements_from_current_reporting",
    "test_benchmark_report_claims.TheBenchmarkComparisonIsRecordedNotGating.test_future_measurements_do_not_overwrite_the_archive",
    "test_related_work.RelatedWorkTests.test_readme_links_to_the_existing_comparison",
    "test_related_work.RelatedWorkTests.test_traceability_comparison_uses_prose_without_rankings_or_promises",
… cut at 3000 characters; the whole file is runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.implement/report.md

</details>

<details><summary><b>26 · supervising → supervising</b> 2026-10-05 22:53Z</summary>

</details>

<details><summary><b>27 · ticket.review-1 · review → completed</b> 2026-10-06 00:25Z</summary>

_findings_ `runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.review-1/findings.md`

```json
{
  "round": 1,
  "verdict": "approve",
  "findings": []
}

28 · ticket.fix-1 · fix → cancelled

29 · ticket.review-2 · review → cancelled

30 · ticket.fix-2 · fix → cancelled

31 · supervising → supervising 2026-10-05 22:53Z

32 · ticket.green · green → completed 2026-10-06 00:33Z · `READY: yes`

green runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.green/green.md

The gate, run 1

Checkout /home/vjovanov/ag/grund/fix/issue-475 on fix/issue-475 at 38cceb5c95; 1 command(s) and the reproducer, 6m56s in all.

grund: the installed grund 0.15.1-dev, since the checkout pins none.

check	command	exit	time
reproducer	bash /home/vjovanov/ag/grund/fix/issue-475/panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh	0	0m00s
1	pre-commit run --all-files	0	6m56s

reproducer. bash /home/vjovanov/ag/grund/fix/issue-475/panta/github-issues-agent-grounds-grund-475-implement-a9ed5e0d/runtime/repro/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/run.sh - passed

Verdict lines:

Expected: Positioning distinguishes supported declarative constraints from severity customization.
Observed PASS docs/roadmap.md: reported misleading passage is absent.
Expected: README links to the existing detailed traceability comparison.
Observed PASS README.md: comparison link found.
0 reported documentation defects found.

1. pre-commit run --all-files - passed

Verdict lines:

grund fmt................................................................Passed
grund init --check (managed block is current)............................Passed
lychee...................................................................Passed
fissile (file size budgets)..............................................Passed
no Claude attribution boilerplate (staged files).........................Passed

READY: yes

33 · supervising → supervising 2026-10-05 22:53Z

34 · ticket.describe · describe → completed 2026-10-06 00:37Z

description runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.describe/description.md

{
  "title": "Clarify factual positioning and performance evidence",
  "body_path": "runtime/supervision/pr-description.md",
  "ticket": "#475",
  "steps_summarized": 17,
  "accounting": "measured",
  "notes": "Body published to draft PR #476 via GitHub REST PATCH after gh pr edit failed on the deprecated GraphQL projectCards field; a fresh API read matched the artifact exactly. Ship should apply the suggested title: the PR still carries the specification-stage title. Readiness, rebase, push, hosted CI and merge remain for ship."
}

pr-description runtime/supervision/pr-description.md

Closes #475

Before, a README reader saw an undated ~722k LoC/s badge and a comparison that ranked tools and promised future coverage parity. Now the badge is retired, and short README links lead to the existing task-oriented comparison, dated benchmark archive and current CI methodology. The existing introduction and section-retrieval examples are reused. This extends §REQ-readme.1.

The new §REQ-readme.evidence requires supported positioning claims, precise shipped boundaries and preserved performance evidence across the README and its supporting documents:

• Related work: replace the ranking matrix and unsupported historical assertions with concrete retrieval, citation-checking, per-file citation-query and declarative-rule tasks. Retained Lychee, OpenFastTrace and Sphinx-Needs claims have a primary-source ledger checked on 2026-10-06, with versions or explicit unversioned status. Correct Lychee's scope to Markdown and HTML. Explain that resolving citations does not prove semantic implementation correctness, and cover does not judge sufficient implementation coverage. Supported chapter/citation constraints are distinct from fixed severity and absent arbitrary executable checks or scripting inside the engine.
• Performance evidence: distinguish the 2026-05-20 local wall-clock run, the historical committed instruction-count snapshot and current generated-fixture comparisons against the PR base branch. Preserve measured tables, raw samples and provenance. Instruction count is a workload-cost proxy under binary/input/toolchain/build/PGO assumptions; the historical unlike-workload timing ratio establishes no general speed ranking. State that instruction-regression limits are reported but unenforced, and acknowledge missing historical compiler/PGO provenance. Future measurement instructions write outside the archive.
• Roadmap: align both positioning milestones and incidental gap-report copy with those facts. Remove coverage-parity promises and unsupported schema-check/interchange policy inferences. Preserve published section coordinates, including work.matrix, and the historical gap proposal's command/design. v2 schema: enforce ordered fields, scalar values, closed shapes, and declaration forms #458 and After 1.0: express place citation and grounding constraints as grounded rule declarations #468 remain proposals.

The committed diff contains four public documents, the README requirement and two existing integration-test files. The requirement and failing tests landed first in 5d0bf1c70b; the documentation correction followed in 38cceb5c95. No new measurements, benchmark machinery, reporting capability, CLI/core/configuration change or product-policy change landed.

Verification at 38cceb5c95:

… cut at 3000 characters; the whole file is runtime/supervision/pr-description.md

35 · supervising → supervising 2026-10-05 22:53Z

36 · ticket.ship · ship → ship-fix 2026-10-06 00:39Z

shipped runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.ship/shipped.md

{
  "pr": 476,
  "url": "https://github.com/agent-grounds/grund/pull/476",
  "branch": "fix/issue-475",
  "merged": true,
  "merge_commit": "f436a6d97613dde5a834b9bf8dd34127530e199e",
  "ci": {
    "cargo test (ubuntu-latest)": "pass",
    "cargo test (macos-latest)": "pass",
    "cargo test (windows-latest)": "pass",
    "commit message attribution gate": "pass",
    "cargo bench (instruction counts)": "pass"
  },
  "fixups": [
    "Changed PR #476 title to “Clarify factual positioning and performance evidence” and verified the returned title and a fresh read."
  ],
  "main": {
    "commit": "f436a6d97613dde5a834b9bf8dd34127530e199e",
    "runs": {
      "CI": "success"
    },
    "status": "green"
  },
  "notes": "Merged as f436a6d97613dde5a834b9bf8dd34127530e199e; main is green on it (1 run(s))."
}

37 · ticket.ship · ship-fix → ship 2026-10-06 00:40Z

fix runtime/ship/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.ship/fix-1.md

The prepare step requested the approved release-title correction. PR #476 still carried the specification-stage title, “Require factual positioning and performance evidence, pinned by tests”. Updated only its title through GitHub REST PATCH and verified it with a fresh REST read. The accepted body, draft status and head 38cceb5 were preserved. The PR remains unmerged; hosted checks and default-branch status have not been assessed in this invocation. No repository change, commit, gate run, push or merge was needed.

FIXUP: Changed PR #476 title to “Clarify factual positioning and performance evidence” and verified the returned title and a fresh read.
SHIP: retry

38 · ticket.ship · ship → awaiting-checks 2026-10-06 00:39Z

39 · ticket.ship · awaiting-checks → merging 2026-10-06 00:40Z

40 · ticket.ship · merging → watching-main 2026-10-06 00:44Z

41 · ticket.ship · watching-main → completed 2026-10-06 00:44Z

Also on disk: runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.plan/proposal.md, runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.evidence.txt, runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.reproduce.py, runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.reproduce.txt, runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.reproducing-upstream.json, runtime/triage/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage.upstream.json

Outcome completed · Verdict: shipped - #476 merged as f436a6d.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage · Validated: kind usability, verdict none, repro yes. The export is runtime/exports/github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.triage/validation.md.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.agora · Planned documentation correction: no disputed ground or cross-family ruling is needed; obtain the owner-reserved editorial direction through clarification and the proposal.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.clarify · The author accepted every assumption on #475 by reaction. The answers are in runtime/clarify/answers.md.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.plan · The author approved the plan on #475 by reaction. Their answer is in runtime/plan/response.md.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.dependencies · Every prerequisite declared for #475 is closed.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.specify · Draft pull request #476 opened from fix/issue-475 with the specification change and the failing test.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.implement · Implemented the approved four-document factual positioning and performance-evidence correction in commit 38cceb5, with source audit and review handoff published.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.review-1 · Approve round 1 with zero findings: 0 blocker, 0 major, 0 minor, 0 nit and 0 adjacent.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.fix-1 · Review round 1 approved with zero findings; no fixes are required before the authoritative green gate.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.review-2 · Review round 1 approved with zero findings and no intervening fix; another review round is unnecessary.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.fix-2 · Review round 1 approved with zero findings; review round 2 is cancelled and no second fix round is needed.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.green · The gate is green: 1 command(s) and the reproducer passed on 38cceb5 in 6m56s.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.describe · Suggested title: “Clarify factual positioning and performance evidence”; the completed description is published to draft PR #476.

Outcome github-issues-agent-grounds-grund-475-implement-a9ed5e0d.ticket.ship · Pull request #476 merged as f436a6d.

Rendered from github-issues-agent-grounds-grund-475-implement-a9ed5e0d/tasks/01-ticket.md by scripts/lifecycle-sync.sh.