agent-release-gate

Run the agent release gate — a portable, wire-level QA harness for the agent runtime. Drives the same product endpoint the playground drives and asserts on the SSE frame stream and real side effects, never on model prose, so it works against any deployment (cloud or self-hosted) from three env vars. Use before an agent-workflows release, or after changing the runner, the SDK agent adapters, the runner Docker images, or the agent service. Triggers: "run the release gate", "QA the agent runtime", "does the agent still work end to end", "pre-release agent QA".

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

63%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The content is highly actionable with excellent validation discipline, but it is a dense 643-line monolith that inlines incident narratives, dated findings, and per-cell trial results that should live in the referenced resources/ files rather than the overview.

Suggestions

Move the inline incident histories, dated findings (2026-08-06/28/30), session UUIDs, and per-trial 2/3 breakdowns into resources/LESSONS.md or coverage.md, leaving the overview as a lean pointer to those files.

Replace the narrative 'Run it' / 'When results lie' prose with a numbered release-gate checklist (export vars → run qa_product.py --all --release-base → run qa_commit_approval.py → run session_control.py when touched → read summary, name every SKIP) so the multi-driver orchestration is one crisp sequence.

Either bundle the referenced resources/ files so the links resolve, or restructure the inlined per-cell descriptions (matrix_w*/l*/b*/c*) into a compact table in coverage.md referenced once from the overview.

Dimension	Reasoning	Score
Conciseness	The 643-line body is noticeably verbose: it inlines extensive incident narratives (AGE-4249, the free-credits 401 of 2026-08-30), per-trial result breakdowns with verbatim model replies, session UUIDs, and time-sensitive dates/PR numbers that the rubric flags unless placed in a deprecated section — none of which is.	2 / 5
Actionability	Fully executable, copy-paste-ready bash commands throughout — env-var exports, `uv run resources/qa_product.py --all --require-store`, and per-journey flags (`--cell`, `--only`, `--burst-size`, `--release-base`) — with specific examples covering the common run shapes.	5 / 5
Workflow Clarity	A clear sequenced flow (set/export env vars → run gate → read results → handle mandatory cells → run companion scripts) with strong explicit validation checkpoints ('Any FAIL blocks the release', SKIP-in-security-areas-is-a-FAILURE, 're-run any prior blocker-level finding after a redeploy'), though the orchestration across qa_product.py, qa_commit_approval.py, session_control.py and the matrix scripts is not presented as one crisp ordered checklist.	4 / 5
Progressive Disclosure	A well-signaled 'Resources (read on demand)' section lists one-level-deep references (coverage.md, LESSONS.md, qa_product.py, matrix_*.py), but no bundle files actually exist in the skill, and large amounts of per-cell lore and findings that belong in those referenced files are inlined directly into the overview.	3 / 5
	Total	14 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is a strong model example: concrete actions, explicit 'when to use' guidance, natural trigger phrases, and a clearly distinct niche, all in third-person imperative voice with no fluff or over-claims.

Dimension	Reasoning	Score
Specificity	Names the domain and multiple concrete actions — 'Run the agent release gate', 'Drives the same product endpoint the playground drives', 'asserts on the SSE frame stream and real side effects, never on model prose' — with comprehensive coverage of what the skill does.	5 / 5
Completeness	Explicitly answers both 'what' (a portable, wire-level QA harness asserting on SSE frames and side effects from three env vars) and 'when' ('Use before an agent-workflows release, or after changing the runner, the SDK agent adapters...') with concrete trigger phrases.	5 / 5
Trigger Term Quality	Lists four natural trigger phrases a user would actually say — "run the release gate", "QA the agent runtime", "does the agent still work end to end", "pre-release agent QA" — covering the common ways this need surfaces.	5 / 5
Distinctiveness Conflict Risk	Occupies a clear niche — agent-runtime release-gate QA — with distinct triggers unlikely to fire for unrelated skills; the wire-level/deployment-agnostic framing further separates it from general QA skills.	5 / 5
	Total	20 / 20 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (644 lines); consider splitting into references/ and linking	Warning
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning

	Total	14 / 16 Passed

Repository: Agenta-AI/agenta
Path: .agents/skills/agent-release-gate/SKILL.md
Commit: e4008fa

Reviewed: 5 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.