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 agentmods add skills/incommensurablehubris/fullstack-director/06-releasenpx skills add IncommensurableHubris/fullstack-director --skill 06-releasegit 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/06-release)<a href="https://agentmods.dev/skills/incommensurablehubris/fullstack-director/06-release"><img src="https://agentmods.dev/badge/skills/incommensurablehubris/fullstack-director/06-release.svg" alt="Measured on agentmods" 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 | $0.00298 | $0.05767 |
| Opus 5 | $0.00149 | $0.02884 |
| Sonnet 5 | $0.00060 | $0.01153 |
| Haiku 4.5 | $0.00030 | $0.00577 |
Grade A, and why
06-release 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 5d 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 — 207 lines — stays where its author put it; the contents beside it link to each section on GitHub.
06 · Release — ship
Two modes. 06-release sprint N ships one verified sprint; 06-release full is the pre-release gate + ship
across all shipped sprints. 06 is the gate the whole spine exists to protect: it consumes 05's
machine-readable verdict tally + the spine's amendment/clarification state and either blocks with every failed
check named and routed, or turns the verified build into a released one — deploy config, one approved plan,
captured evidence, health + smoke verification, an auditable release report. Your graded value is not "can run
deploy commands" (a strong operator does that too) — it is the machine-checkable protection-rule gate plus the
evidence-verified, human-approved, no-secrets release record.
Operating principle — gate on recorded state, deploy only an approved plan, verify before traffic
- Gate first, mechanically. Every check reads recorded machine state — the qa-report frontmatter tally,
amendment-log.jsondispositions, a marker scan, git identity — never re-derives judgment: no code re-review, no suite re-run (05already executed the oracles at the exact commit the identity check pins). Fail-closed: missing or unparseable state is a BLOCK, never an assumption. Evaluate all checks; a blocked release names every failed check with its routed next command, and still writes the release report (status: BLOCKED). - Plan/apply with ONE human approval. Write the deploy plan — exact commands, expected evidence per step, blast
radius, and the rollback path (stated before approval, or the plan is invalid) — and pause for a single
approval. Only then execute, capturing
command · exit code · outputper step. No irreversible external action (push, deploy, DNS, tag-push) before the approval; no per-step re-confirmation nagging either. - Verify before traffic; roll back on failure. Deploy ≠ release: health-check the new version (liveness +
readiness) before it takes traffic where the platform allows, then switch, then smoke the live critical flows —
captured output, not assertions. A failed verification executes the pre-stated rollback and re-verifies the
restored version →
ROLLED-BACK. Every failure or near-miss becomes a guardrail row.
What ships with it
34 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- evals/build_fixture.py 8.9 KB runs code
- evals/check_release.py 42 KB runs code
- evals/evals.json 17 KB
- evals/fixtures/app/.gitignore 195 B
- evals/fixtures/app/scripts/deploy.js 1.1 KB runs code
- evals/fixtures/app/scripts/health.js 1.1 KB runs code
- evals/fixtures/app/scripts/smoke.js 1.7 KB runs code
- evals/fixtures/app/src/digest.js 1.0 KB runs code
- evals/fixtures/app/test/digest.test.js 1.8 KB runs code
- evals/fixtures/base/docs/architecture/adr/ADR-001.md 1.4 KB
- evals/fixtures/base/docs/architecture/adr/README.md 334 B
- evals/fixtures/base/docs/architecture/system.md 4.0 KB
- evals/fixtures/base/docs/planning/sprints/sprint-01.md 2.1 KB
- evals/fixtures/base/docs/spec/amendment-log.json 668 B
- evals/fixtures/base/docs/spec/architecture-constraints.md 1.3 KB
- evals/fixtures/base/docs/spec/capabilities/digest.md 2.1 KB
- evals/fixtures/base/docs/spec/capabilities/standups.md 2.2 KB
- evals/fixtures/base/docs/spec/specification.md 2.9 KB
- evals/fixtures/cases/blocked-spine/docs/spec/amendment-log.json 997 B
- evals/fixtures/cases/blocked-spine/docs/spec/capabilities/digest.md 2.1 KB
- evals/fixtures/cases/blocked-spine/qa-report.tmpl.md 3.6 KB
- evals/fixtures/cases/blocked-verdict/qa-report.tmpl.md 4.4 KB
- evals/fixtures/cases/blocked-verdict/src/digest.js 1.4 KB runs code
- evals/fixtures/cases/blocked-verdict/test/digest.test.js 1.5 KB runs code
- evals/fixtures/cases/blocked-verdict/test/review/req-008-grouping.test.js 770 B runs code
- evals/fixtures/cases/clean/qa-report.tmpl.md 3.2 KB
- evals/fixtures/cases/patch/docs/planning/backlog.md 1.3 KB
- evals/fixtures/cases/patch/docs/planning/patches/patch-001.md 1.3 KB
- evals/fixtures/cases/patch/qa-report.tmpl.md 3.0 KB
- evals/README.md 15 KB
- references/deploy-verification.md 12 KB
- references/release-gate.md 22 KB
- templates/deployment-config.md 4.8 KB
- templates/release-report.md 8.5 KB
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.
- 5d ago First seen · 207 lines · 298 tokens per session scan A 240097cb58bf
06-release is a skill published in the GitHub repository IncommensurableHubris/fullstack-director (5 stars, last pushed 25d ago), licensed MIT. It adds 298 tokens to every session and 5,767 once invoked, about $0.0015 per session on Opus 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
land-and-deploy
Land and deploy workflow. (gstack).
publish-site
Versioned site deploys to GitHub/Cloudflare/Netlify Pages.
gsd-complete-milestone
Archive completed milestone and prepare for next version.
release
Release workflow for deploying OpenEnv environments to Hugging Face Spaces and keeping canonical references in sync.
release-archivist
Close out a spec-superflow change with verification, summary, and archive readiness. Invoke when implementation is complete, verification is underway, or the user asks for a final wrap-up.
slack-announcement
Produce a themed Slack post announcing Ark releases. Pulls from BOTH the Ark core repo (mckinsey/agents-at-scale-ark) and the Marketplace repo (mckinsey/agents-at-scale-marketplace), validates docs links, groups features by theme with relevant emoji, and ends with a "Coming next" section sourced from the current…