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 ayberkcansever/ai-skills --skill retro-buildgit clone --depth 1 https://github.com/ayberkcansever/ai-skillsWrote 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/ayberkcansever/ai-skills/retro-build)<a href="https://agentmods.dev/skills/ayberkcansever/ai-skills/retro-build"><img src="https://agentmods.dev/badge/skills/ayberkcansever/ai-skills/retro-build/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/ayberkcansever/ai-skills/retro-build"><img src="https://agentmods.dev/badge/skills/ayberkcansever/ai-skills/retro-build.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00053 | $0.01117 |
| Opus 5.5 | $0.00021 | $0.00447 |
| Sonnet 5.5 | $0.00011 | $0.00223 |
| Haiku 4.5 | $0.00005 | $0.00112 |
Grade A, and why
retro-build 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 14d 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.
How it starts
The opening of the file, as written. The whole thing — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Retro Build
Input: $ARGUMENTS
Judge the process, not the code. Never edit a skill file without per-item
approval. The skills live in <skills> =
$(cd -P "${CLAUDE_SKILL_DIR}/.." && pwd) — resolved, because the install
is per-folder symlinks. It must be inside a git repo so every change is a
revertible commit with its evidence — a copied install is not, so ask for
the skills repo path. retro-log.md is <skills>/retro-log.md; create it
on the first retro.
Reads: the plan's spec and plan, retro-log.md, git log --grep 'retro('.
Writes: approved skill edits, retro-log.md, and repo facts into
docs/quirks.md or AGENTS.md.
1. Collect
Read the plan and spec in <root>/docs/plans/<plan>/ (<root> = the main
checkout; the plan named in the input, or the
plan from this session). Missing both → "nothing to retro", stop. Quote each
signal verbatim with its location: > Drift: notes, ## Blockers entries, R findings from
cycles > 1, (overrides recommendation: …), supersedes D<n>, ## E2E
fails, ## Assumptions entries proven wrong, accepted risks that
materialized, failed ## Release steps, and any escaped production defect.
Print:
metrics(<plan>): lane=<l> tasks=<n> gate-first-pass=<k>/<n> review-cycles=<c> drift=<d> blockers=<b> supersedes=<s> e2e-fail=<e> assumed-wrong=<a>
2. Attribute
- Behaviour the spec never mentioned, a late-surfacing constraint, an overridden recommendation, a wrong assumption, or a symbol / path the plan got wrong that the repo could have answered → prepare-plan.
- Preflight or environment blocker, or a gate cap hit on a runnable command → implement-plan.
- Finding first caught in cycle > 1, or an escaped defect → review-build.
- E2E fail → prepare-plan if no decision covered the behaviour, else review-build.
- Genuinely unknowable environment state → no node, skip.
- Recurring domain fact (filter semantics, scoping, timezone, idempotency) →
the repo's quirks doc. Repo tooling → that repo's
AGENTS.md.
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.
- 14d ago First seen · 99 lines · 53 tokens per session scan A eb493cbd7380
retro-build is a skill published in the GitHub repository ayberkcansever/ai-skills (2 stars, last pushed 14d ago), licensed MIT. It adds 53 tokens to every session and 1,117 once invoked, about $0.0002 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 2026-10-03.
Other skills, from other repositories
req-create
Create structured requirement documents from user ideas. Use when capturing new features, tasks, or user stories.
pn-create-workflow-roadmap
Generate a project workflow roadmap (docs/WORKFLOW.md) mapping plan phases to pn-core commands with model tier recommendations and per-step token cost estimates. Runs after pn-writing-plans.
pn-program-orchestration
Conductor for featureprogram: split ≥2 vertical slices, DAG, merge queue. Use when the user asks for parallel slices or /pn-program — not for a single-slice feature.
pn-github-vertical-slices
Break a plan or PRD into tracer-bullet vertical slices and create GitHub Issues via GitHub MCP (official github/github-mcp-server). Dependencies first (AFK/HITL). Use after pn-writing-plans or pn-create-prd when work must ship as Issues.
pn-paperclip
Interact with the Paperclip control plane API to manage tasks, coordinate with other agents, and follow company governance. Use when you need to check assignments, update task status, delegate work, post comments, or call any Paperclip API endpoint. Do NOT use for the actual domain work itself (writing code, research…
pn-feedback-analysis
Collect, analyze, and synthesize user feedback into actionable product insights. Themes, prioritization, voice of customer.