[pithead-developer]

Closes #3197

The README overstated privacy, rollback and rig-control guarantees. It now matches the linked guides: Tor-first defaults with documented exceptions, variable PPLNS and solo Tari payouts, wallet-free miner configuration scoped to Pithead workers, token-authenticated adopted RigForge control, OS-only A/B fallback, and the three hardware arrangements.

Sources checked: docs/privacy.md:10, docs/dashboard.md:341, docs/workers.md:17, docs/dashboard.md:431, docs/appliance.md:398, docs/appliance.md:451, CHANGELOG.md:26, and docs/hardware.md:10. Each corrected README claim links to its guide. Latest-release links and the deployment chooser remain outside this issue.

Ponytail scope review: PASS; the diff contains only the five requested README corrections and their source links, with nothing to cut. Independent verifier and security reviewer: PASS, no findings; their reviewed diff is the README-only change committed as 658dd8d4277f248a73f04c82370780d41ca6bbfe.

Validation:

• git diff --check: PASS.
• make lint-topology: PASS, including its selftests.
• make lint-docs-voice lint-path-references lint-file-budget lint-pithead-build lint-md: exit 2. The first four targets passed, including their selftests; the final lint-md target could not start because local npx is unavailable (exit 127). Markdownlint remains a required GitHub CI check; no check was weakened or skipped to claim a local pass.
• gh pr checks 3205 --repo p2pool-starter-stack/pithead: current-head checks are pending, with no failures reported at the first inspection. The two shipped-image CVE jobs report skipping. The runner will gate review/merge on completed required checks.

Proof follows the docs row in docs/dev/testing-strategy.md:53: GitHub tiers 1 to 3 provide the applicable checks. No runtime behavior changed, so no runtime tests were added and no tier-4 bench job was submitted. Runtime routing, payouts, rig controls and appliance fallback were not exercised by this prose correction. The repository's external link checker is scheduled, not a per-PR check; new local links and heading anchors were reviewed against the guides.

[pithead-developer]

Ready for review

At 658dd8d4277f248a73f04c82370780d41ca6bbfe, the README covers all five requested claim corrections and links their canonical guides. Both the adversarial verifier and security reviewer returned PASS with no findings; scope review found nothing outside the issue to cut.

Local docs voice, path references, topology, file budget and CLI build checks passed. The combined documentation command then failed at markdownlint because local npx is unavailable; GitHub CI supplies that required check. Current-head GitHub checks are pending with no failures reported at the first inspection, so this draft still needs those completed results and the runner’s independent review before merge.

This changes prose only. No routing, payout, rig-control or appliance runtime test was run, and no tier-4 job is required by the documentation proof row. New local links and anchors were reviewed; the external link check runs on its scheduled workflow.

[pithead-reviewer]

Adversarial review: PASS

PASS at 658dd8d: the README now qualifies Tor routing and OS fallback, explains payout variance and Pithead-only wallet configuration, states adopted RigForge control prerequisites, and names all three hardware roles with canonical guide links. Current-head GitHub checks and local documentation, topology, file-budget, and CLI build checks pass; no runtime files changed.

Acceptance verified against the owner’s #3197 and the developer App’s latest Ready for review. This is a README-only correction; no runtime, integration, Docker, or bench tier was run. Security review found no new secrets, exposed services, injection surface, or supply-chain changes. Ponytail review: lean already; no complexity findings.

Evidence

• pwd; git status --short; git rev-parse HEAD; git remote -v; cat AGENTS.md: clean checkout at 658dd8d; origin is p2pool-starter-stack/pithead. Local checkout location retained privately.
• Read graphify, ponytail, and ponytail-review SKILL.md: applied the diff complexity review; no graph existed. No branch files were edited.
• gh pr view 3205 --repo p2pool-starter-stack/pithead --json title,body,author,baseRefName,headRefName,headRefOid,isDraft,comments,reviews,commits,files: develop base; one README-only commit; developer App's Ready for review dated 2026-10-06T15:40:12Z; no earlier RETURN or unanswered review.
• gh issue view 3197 --repo p2pool-starter-stack/pithead --json title,body,author,state,comments: owner's five corrections and explicit exclusions read; no comments.
• Read docs/dev/ai-workflow.md, CONTRIBUTING.md, docs/dev/testing-strategy.md, docs/dev/STYLE.md, and docs/dev/repo-map.md: docs-only proof row applies; Docker-dependent make test is not permitted on this Mac and was not run.
• gh issue view 3180 and 3182 --repo p2pool-starter-stack/pithead --json title,body,author,state,comments: direct linked context read; umbrella release proof does not add runtime scope to docs: align README claims with the guides before 2.0.0 (#3180 sweep) #3197. No recursive evidence tracing.
• gh api repos/p2pool-starter-stack/website/issues/42 and its comments: HTTP 404 for the shorthand repository. Resolved website#42 to p2pool-starter-stack.github.io/pull/42, whose owner-authored body was available.
• gh api repos/Correct overstated support, privacy, payout and setup claims p2pool-starter-stack.github.io#42 and /comments; gh pr view 42 --repo p2pool-starter-stack/p2pool-starter-stack.github.io --json state,comments,reviews: corrections merged; no comments or reviews.
• gh api repos/Align org profile support, privacy, fee and resource claims with the docs .github#10 and /comments; gh pr view 10 --repo p2pool-starter-stack/.github --json title,body,author,state,comments,reviews: owner-authored org corrections merged; no comments or reviews. These completed prose PRs are context, not a RigForge runtime companion.
• git diff origin/develop...HEAD -- README.md; nl -ba README.md: 17 additions, 13 deletions in README.md only. Existing overclaims corrected in place; no relaxed tests, swallowed errors, hardcoded returns, TODO fixes, or silently narrowed scope. Ponytail review found no machinery to cut.
• Numbered canonical-guide reads: README.md:17-18 and :41-44 match docs/privacy.md:10-20 and :26-34; wallet scope at README.md:45-47 matches docs/workers.md:17-18 and :22-43.
• Numbered canonical-guide reads: payout variance at README.md:37-38 matches docs/dashboard.md:317-343; control prerequisites at README.md:51-56 match docs/dashboard.md:394-445 and docs/workers.md:410-416. Plain XMRig monitoring remains supported.
• Numbered canonical-guide reads: OS-only fallback at README.md:89 and :151 matches docs/appliance.md:398-409 and :451-452 plus CHANGELOG.md:20-26; the three hardware roles at README.md:153 match docs/hardware.md:6-18. No changed behavior requires a new unit test or a without-fix mutation run.
• Read Makefile lint targets and .github/workflows/ci.yml schedule conditions: shipped-image sweep and tracking jobs are schedule-only; their PR skips are expected.
• gh pr checks 3205 --repo p2pool-starter-stack/pithead, then --json name,state,bucket summary: 39 checks, 37 pass, two schedule-only skips, no failing or pending checks.
• gh api repos/p2pool-starter-stack/pithead/actions/runs/{37489260673,37489260397,37489260143,37489254152}: each completed with success and head_sha 658dd8d; CI, Shell, Tari config parse, and CodeQL respectively.
• gh api repos/p2pool-starter-stack/pithead/actions/jobs/112357348470 and /112357348538: current-head per-surface lint, dashboard coverage, patch coverage, and fake-daemon steps all success. No quoted bench-ci job exists or is required for this README-only diff; no bench tier was run.
• git diff --check origin/develop...HEAD: exit 0, no whitespace errors.
• make lint-topology: passed all selftests and repository scan; no prohibited topology.
• make lint-docs-voice lint-path-references lint-file-budget lint-pithead-build lint-md: exit 0; voice clean in 75 prose docs; all named paths resolve, 790 source targets checked and 303 dynamic expressions explicitly skipped; file-budget ratchet and CLI build/order selftests pass; pinned markdownlint-cli2 0.18.1 checked 81 files with zero errors.
• python3 read-only canonical-guide link/anchor assertions: exit 0; ten distinct README guide destinations resolve, including every changed heading anchor.
• python3 runner detector over captured PR body, issue comments, review bodies, PR inline comments, and docs: align README claims with the guides before 2.0.0 (#3180 sweep) #3197 comments: all scans exit 0, no hits. Initial --help attempt is unsupported and raised FileNotFoundError; actual file scans subsequently succeeded.
• python3 runner detector over directly linked 2.0.0 launch claims: RC evidence, Tor tradeoffs and RigForge comparison #3180, Guide new users from deployment choice to their first accepted mining work #3182, website#42, and .github#10 bodies/comments/reviews: all four scans exit 0, no historical topology notes.
• gh auth status: active credential is pithead-reviewer[bot]; owner credential inactive. Credential values not published.
• git diff --stat origin/develop...HEAD; git log origin/develop..HEAD --format='%H %s'; git status --short; git rev-parse HEAD; gh pr view 3205 --json headRefOid,baseRefName,isDraft: sole README correction commit, unchanged clean checkout and matching remote head; develop base. Draft status does not affect the verdict.

Non-blocking note

• PR docs: align README claims with canonical guides #3205 body:13: Replace the stale checks-pending entry with the completed current-head result: 37 passing checks and two expected schedule-only CVE skips. The review also ran the pinned local markdownlint successfully (81 files, zero errors).

Publication: the public review was topology-scanned and returned clean. A session attempt to write the adversarial-review status was refused with exit 3: sessions do not write commit statuses or check runs (the runner posts the review verdict). The runner must publish the status from this verdict object; the session wrote no status and performed no merge.

[pithead-reviewer]

adversarial-review: PASS at 658dd8d, round 1, by the pithead-reviewer reviewer (a fresh session that did not produce this head; gpt-6.1-sol).

PASS at 658dd8d: all five README corrections match their canonical guides, with no runtime changes or scope findings. Current-head checks and local documentation gates pass; the verdict and evidence are published on the PR.

Evidence

• git status --short; git rev-parse HEAD; git remote -v; cat AGENTS.md: clean checkout at 658dd8d in the correct repository.
• Read graphify, ponytail, and ponytail-review skills: no existing graph; completed the diff complexity review without editing branch files.
• gh pr view 3205 --repo p2pool-starter-stack/pithead --json title,body,author,baseRefName,headRefName,headRefOid,isDraft,comments,reviews,commits,files: develop base, one README-only commit, latest developer App Ready for review read; no prior RETURN.
• gh issue view 3197 --repo p2pool-starter-stack/pithead --json title,body,author,state,comments: owner's acceptance requires qualified Tor scope, OS-only fallback, payout variance and Pithead-only wallet configuration, adopted RigForge control prerequisites, and three hardware roles. Release links and the deployment chooser are explicitly excluded.
• Read repository workflow, contributing, testing strategy, documentation style, and repo map: documentation proof row applies. Docker-dependent make test was not run under the explicit Mac restriction.
• gh issue view 3180 and 3182 --repo p2pool-starter-stack/pithead: direct linked bodies and comments read; broader release and chooser work does not expand docs: align README claims with the guides before 2.0.0 (#3180 sweep) #3197.
• gh api repos/p2pool-starter-stack/website/issues/42 and /comments: shorthand lookup returned HTTP 404. Canonical p2pool-starter-stack.github.io/pull/42 was subsequently available.
• gh api repos/Correct overstated support, privacy, payout and setup claims p2pool-starter-stack.github.io#42 and /comments; gh pr view 42 in that repository: owner-authored correction merged, no comments or reviews.
• gh api repos/Align org profile support, privacy, fee and resource claims with the docs .github#10 and /comments; gh pr view 10 in that repository: owner-authored correction merged, no comments or reviews. No runtime companion PR requires review or merge ordering.
• git diff origin/develop...HEAD -- README.md; numbered README inspection: 17 additions and 13 deletions, all five corrections made in place. No weakened tests, swallowed errors, hardcoded returns, TODO fixes, or silently narrowed scope. Ponytail review found nothing to cut.
• Numbered canonical-guide inspection: README.md:17 and :41 match docs/privacy.md:10 and :26; README.md:45 matches docs/workers.md:17 and :22.
• Numbered canonical-guide inspection: README.md:37 matches docs/dashboard.md:317 and :341; README.md:51 matches docs/dashboard.md:394 and :431 plus docs/workers.md:410. Plain XMRig monitoring remains supported.
• Numbered canonical-guide inspection: README.md:89 and :151 match docs/appliance.md:398 and :451 plus CHANGELOG.md:26; README.md:153 matches docs/hardware.md:6. Security review found no added secrets, exposed services, injection surface, or supply-chain changes.
• Read Makefile lint recipes and CI workflow conditions: both skipped shipped-image CVE jobs run only on schedule.
• gh pr checks 3205 --repo p2pool-starter-stack/pithead; JSON check summary: 37 passing checks, two expected schedule-only skips, no pending or failing checks.
• gh api actions/runs/37489260673, /37489260397, /37489260143, and /37489254152 in pithead: CI, Shell, Tari config parse, and CodeQL all completed successfully at the exact reviewed head.
• gh api actions/jobs/112357348470 and /112357348538 in pithead: current-head lint, dashboard coverage, patch coverage, and fake-daemon steps succeeded. No bench job is quoted or required for the README-only paths.
• git diff --check origin/develop...HEAD: exit 0, no whitespace errors.
• make lint-topology: selftests and repository scan passed.
• make lint-docs-voice lint-path-references lint-file-budget lint-pithead-build lint-md: exit 0. Voice clean in 75 prose docs; paths resolve; 790 source targets checked with 303 dynamic expressions explicitly excluded. File-budget and CLI build/order gates passed. Pinned markdownlint checked 81 files with zero errors.
• python3 canonical-guide link and anchor assertions: exit 0; ten distinct README guide destinations resolve, including all changed heading anchors.
• python3 runner detector over captured PR body, PR comments, reviews, inline comments, and docs: align README claims with the guides before 2.0.0 (#3180 sweep) #3197 comments: all scans exit 0, no hits. Initial unsupported --help invocation raised FileNotFoundError; subsequent actual scans succeeded.
• python3 runner detector over directly linked 2.0.0 launch claims: RC evidence, Tor tradeoffs and RigForge comparison #3180, Guide new users from deployment choice to their first accepted mining work #3182, website#42, and .github#10: all four scans exit 0, no historical topology hits.
• gh auth status: active identity pithead-reviewer[bot]; owner credential inactive.
• git diff --stat origin/develop...HEAD; git log origin/develop..HEAD; git status --short; git rev-parse HEAD; gh pr view 3205 --json headRefOid,baseRefName,isDraft: sole README correction commit, matching remote head, clean tree. Draft status does not affect PASS.

Findings

• [note] PR docs: align README claims with canonical guides #3205 body:13: Refresh the stale checks-pending entry: the reviewed head has 37 passing checks and two expected schedule-only CVE skips. Local pinned markdownlint also passed over 81 files with zero errors.