Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add doodledood/manifest-dev --skill review-pr-holisticgit clone --depth 1 https://github.com/doodledood/manifest-devWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/doodledood/manifest-dev/review-pr-holistic)<a href="https://agentmods.dev/skills/doodledood/manifest-dev/review-pr-holistic"><img src="https://agentmods.dev/badge/skills/doodledood/manifest-dev/review-pr-holistic/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/doodledood/manifest-dev/review-pr-holistic"><img src="https://agentmods.dev/badge/skills/doodledood/manifest-dev/review-pr-holistic.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00074 | $0.00812 |
| Opus 5.5 | $0.00030 | $0.00325 |
| Sonnet 5.5 | $0.00015 | $0.00162 |
| Haiku 4.5 | $0.00007 | $0.00081 |
Grade A, and why
review-pr-holistic scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 20d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
review-pr-holistic
Turn a pile of findings into the comments a careful reviewer would actually post. This skill edits and consolidates; it generates nothing new and posts nothing itself.
Receives. Fleet findings — severity-graded defect findings from the reviewer dimensions, with dimension and severity retained, Low findings filtered by the caller except those from defect-class — and judgment findings — a distinct non-severity class, each { trigger, concrete evidence, author-facing question }, already synthesized to one question per root; the PR history (all comments and threads on the PR, including ours from any prior pass, the author's recent commit messages on the branch, the PR description); bundle context for each linked PR (diff, description, top-level conversation — no inline review comments from linked PRs); the manifest if present; the reviewed range for this invocation — the code the fleet's defect findings are bounded to; the caller's voice profile — its rules for how a posted comment reads; and any truncation the caller did, to carry forward.
Does, in this order:
- Prunes duplicate concerns already covered by existing threads, findings refuted by evidence, and defect findings whose entire objection is to an explicit intentional requirement. A stated intention establishes scope; evidence establishes behavior. Retain a premise question unless the PR already answers that specific concern. An author's claim alone does not refute a demonstrated failure.
- Dedupes across reviewers: merge near-duplicates into one comment when they raise the same underlying concern.
- Premise-subsumption. When a judgment finding questions the necessity or existence of a surface, subordinate the fleet's findings on that same surface under the premise question — drop the low-value ones, and fold any that survive beneath it ("if you keep X: also a, b") rather than listing them as independent peers. Retain a defect-class finding, including Low, when its remediation obligation remains unmet; a non-blocking premise question cannot discharge it. Never hard-delete a severe independent defect: the judgment finding is non-blocking, so a defect that matters even if the surface stays is still surfaced on its own.
- Bounds surfaced defect findings to issues introduced or exposed by the reviewed range, unless the range is the full PR diff — which covers a first review and a repeat review of an unchanged head alike. This keys on the range, never on whether we have reviewed before: bounding a repeat pass to the delta since our last review would have it read the whole PR and then surface nothing. A repair in the reviewed range exposes its enabling boundary and affected sites even when their lines are unchanged; an unrelated later delta does not reopen the old repair. Judgment findings are exempt: they are whole-PR by construction and stay exempt from the drop-Low step too — they surface whenever their evidence fired and the PR does not already cover the point.
- Anchors each surviving finding to exactly one of inline file:line (default), file-level (a whole-file concern), or PR-level (cross-cutting, no specific anchor).
- Rewrites every comment body and any drafted reply in the supplied voice profile.
- Omits a summary header by default — adds one only when there is a real overall take the per-comment list misses (one short sentence, voice-compliant, no boilerplate).
Returns. The comments to post (anchor plus voice-compliant body), the summary header text if any, truncation notes, and a brief dropped-findings tally with the dominant reasons.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 20d ago Changed a4279ffa5710
- 1mo ago Changed 9a76ef2b4cca
- 1mo ago First seen · 24 lines · 74 tokens per session scan A 7b21b2d07cef
review-pr-holistic is a skill published in the GitHub repository doodledood/manifest-dev (75 stars, last pushed 2d ago), licensed MIT. It adds 74 tokens to every session and 812 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
great_cto
Use when the CTO describes a feature, task, or project goal. Orchestrates the full SDLC pipeline automatically based on project type.
opportunity-solution-tree
Build an Opportunity Solution Tree (OST) to structure product discovery — map a desired outcome to customer opportunities, possible solutions, and experiments. Based on Teresa Torres' Continuous Discovery Habits. Use when the team is unclear what to build next, when multiple opportunities compete, or before writing a…
skeptical-triage
Reusable 3-round self-challenge + arbiter pattern for filtering false positives from findings/verdicts. Use when the cost of a false-positive gate block exceeds the cost of 4 extra LLM turns.
anti-patterns
Catalogue of known SDLC anti-patterns that greatcto agents must actively reject when reviewing architecture, plans, code, or post-mortems. Used by architect (pre-impl), pm (planning), senior-dev (impl), l3-support (post-incident).
well-architected
6-pillar architecture review framework. Adapted from AWS Well-Architected for use by greatcto's architect agent on every non-nano ARCH document. Forces explicit answers across operational excellence, security, reliability, performance, cost, and sustainability — not just feature design.
product-economics
Does this product make money at a price someone will pay? Forces contribution margin, a price with a stated basis, and a bottom-up market size — each number labelled measured / assumed / unknown, so a guess can never be read as a calculation.