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 IncommensurableHubris/fullstack-director --skill feedbackgit clone --depth 1 https://github.com/IncommensurableHubris/fullstack-directorWrote 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/incommensurablehubris/fullstack-director/feedback)<a href="https://agentmods.dev/skills/incommensurablehubris/fullstack-director/feedback"><img src="https://agentmods.dev/badge/skills/incommensurablehubris/fullstack-director/feedback/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/incommensurablehubris/fullstack-director/feedback"><img src="https://agentmods.dev/badge/skills/incommensurablehubris/fullstack-director/feedback.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 | $0.00201 | $0.01247 |
| Opus 5.5 | $0.00080 | $0.00499 |
| Sonnet 5.5 | $0.00040 | $0.00249 |
| Haiku 4.5 | $0.00020 | $0.00125 |
Grade A, and why
feedback 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 1mo 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
feedback · Capture framework friction — a sixty-second detour, then back to work
A cross-cutting activity (prefix-less, like status): the capture half of the framework's
dogfood→distill→upstream loop (shared/feedback-loop.md — read it once; this skill is its clerical arm). You run
in a consumer repo. The finding you write here is triaged later in the framework's master repo — you never fix
vendored files yourself (upstream-first).
The one routing test — before writing anything
Where would the fix land?
- In the vendored framework (
.agents/skills/**,.claude/skills/**,shared/**,tools/vendor.py, a gate/template/protocol) → this skill. Continue below. - In this project's own code or realizations → a product bug: the normal chain owns it (
/05-reviewerfindings →/04-builder), not the ledger. - In this project's spine (a requirement/constraint is wrong) → the amendment protocol
(
shared/spec-amendment-protocol.md), not the ledger. - Genuinely unsure → ledger it (
severity: friction-only, say so inactual:) — a mis-filed entry costs one triage minute; a lost one costs the fix.
The procedure (do all five, in order)
- Stamp. Read
source_commitfrom.director/vendor-manifest.json. Absent/unreadable → useunknown (pre-manifest vendor)and note that a re-sync will fix stamping. - Number. Read
docs/framework-feedback.md; next id =max(FB-NNN) + 1, zero-padded (none yet →FB-001). If the ledger file itself is absent, create it first from the template inshared/feedback-loop.md§1. - Fill the entry — every field, per the ledger's own template:
framework:(the SHA from step 1) ·harness:(the one you are running on:claude-code | codex | gemini-cli) ·seat:(00-discovery…status, orshared/vendor) ·severity:(blocks-gate — a gate/verdict was wrong or unreachable · wrong-but-recoverable — incorrect output you corrected · friction-only — worked, but cost real time) ·expected:/actual:(the load-bearing pair — a mood is not a finding) ·repro:(minimal steps or the exact command) ·artifact:(the transcript / diff / report path that shows it) ·status: open. - Append the entry to the end of
docs/framework-feedback.md. Append-only: never edit or renumber an existing entry (supersede with a new one that names it), never compact the ledger. (2026-07-28) Thestatus:line is the one exception, and it is not yours to write. The master repo sets it at sync time —resolved-in <sha>, ordeferred-in <sha> (trigger: …)when triage acknowledged the entry but did not action it, sometimes with alanded-as:line when the fix differed from what you proposed. So: adeferred-inentry is not ignored — keep your workaround until its stated trigger fires; and an entry still readingopenafter a sync means it was not disposed, which is worth raising. Never set your own status beyond the initialopen. - Confirm + resume. Tell the user the id (
logged FB-007) in one line and return to the interrupted work — this skill is a detour, never a takeover. If you hot-patched a vendored file to unblock (the emergency escape hatch), the entry isblocks-gateandactual:describes the patch.
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.
- 1mo ago First seen · 61 lines · 201 tokens per session scan A 76fccba1afd1
feedback is a skill published in the GitHub repository IncommensurableHubris/fullstack-director (5 stars, last pushed 2mo ago), licensed MIT. It adds 201 tokens to every session and 1,247 once invoked, about $0.0008 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 2026-08-31.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
diagnosing-superpowers
Use when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session…
gsd-forensics
Post-mortem investigation for failed GSD workflows — diagnoses what went wrong.
gsd-audit-fix
Autonomous audit-to-fix pipeline — find issues, classify, fix, test, commit.
gsd-ns-review
Invoke this Pi skill with /skill:gsd-ns-review.
kiro-debug
Investigate implementation failures using root-cause-first debugging. Use when an implementer is blocked, verification fails, or repeated remediation does not converge.