release-gate 0.9.4

AI agent release decision engine - readiness scoring, regression gate, eval runner, trace validation, and evidence packs for CI/CD.

release-gate

The pre-deploy release gate for AI agents. It renders an evidence-based PROMOTE / HOLD / BLOCK verdict — catching the agent-layer risks that SAST, guardrails, and evaluators structurally miss.

v0.9.4 — a lean, three-dependency CLI (pip install release-gate no longer pulls a web/SaaS stack) and a reproducible 79-case benchmark that covers every rule (≥2 vulnerable + ≥2 clean look-alikes each), so the zero-false-positive claim can be checked, not just read. Both sit on top of the v0.9.0 agent-safety catalog (9 new rules + 2 precision upgrades), holding the precision bar at 0 false positives on that labeled benchmark and a framework dogfood (llama_index / crewAI / langgraph / open-interpreter): indirect prompt injection from RAG/tool/HTTP provenance (RG-PROMPT-002), model-driven SSRF / filesystem / SQL sinks (RG-ACTION-002/003/004), secret/PII → prompt data-egress to the provider (RG-SECRET-002, an agent-aware egress path conventional SAST lacks context to model), taint-aware deserialization (RG-EXEC-004), unvalidated model-output parses (RG-PARSE-001), and tool blast-radius + irreversibility gates (RG-TOOL-001 / RG-GATE-001) — plus confirmed taint through the canonical resp.choices[0].message.content extraction and a reproducible PR-gate demo. See the catalog below. Builds on 0.8.5's release-gate pr, the AI-change review gate: one PROMOTE/HOLD/BLOCK on what a pull request introduced (net-new agent risk + lockfile/behaviour drift), plus a GitHub Action command: pr; 0.8.4's security-hardened MCP server (pip install 'release-gate[mcp]'); and 0.8.0–0.8.2's AST-based evidence-citing analysis, deserialization calibration, and team-adoption workflow (--mode / --baseline / --pr-comment).

Why it's not SonarQube: a SAST tool sees eval(x) and asks "is x tainted by SQL/HTTP?" — it has no concept of "x is the model's reply." That blind spot is the entire agent layer: eval/pickle of model output (the CVE-2025-51472 RCE class), user input reaching a system prompt, LLM loops with no cost ceiling. Guardrails filter one input; evaluators score one output; neither blocks a release. release-gate is the gate.

Try it in 30 seconds

pip install release-gate

# ── The wedge: gate a pull request on what IT introduced ──
# One PROMOTE / HOLD / BLOCK on net-new agent risk only (inherited debt is
# shown, never gated) + prompt/model drift. Runs in CI on the PR branch.
git checkout my-feature-branch
release-gate pr --base origin/main
release-gate pr --base origin/main --comment   # GitHub-ready PR comment

# ── Or audit a whole repo (the broader lens) ──
release-gate audit . --mode ci                                    # your repo, in CI
release-gate audit https://github.com/org/repo --mode public-advisory  # any public repo, advisory

Lean by design. pip install release-gate pulls three small, well-audited libraries — pyyaml, jsonschema, cryptography — and nothing else. No web framework, no database driver, no auth stack in the CLI's dependency tree. The release-gate.com server stack is an opt-in extra (pip install 'release-gate[api]'), and the MCP server is another ('release-gate[mcp]').

Output:

  Repo    https://github.com/your-org/your-ai-agent
  Agents  OpenAI / Agents SDK (4 files), LangChain (12 files)

  Readiness Score   42 / 100   ████░░░░░░

  Agent Code Safety  28/100  BLOCK   4 high · 18 med · 0 low
     Driving the score: Dangerous execution sink ×4; LLM call with no token ceiling ×18
  Governance         50/100  Partial   4/8 safeguards declared

  Decision:  ✗  BLOCK

Two axes, on purpose:

• Agent Code Safety — an objective score from the code itself: prompt-injection surfaces, exec/shell sinks fed by model output, LLM calls with no token ceiling, hardcoded keys. It moves per repo and doesn't depend on adopting anything. These are the agent-layer risks generic SAST/SonarQube don't model — release-gate is the layer on top, not a replacement.
• Governance — maturity of your declared, enforceable safeguards (budget ceiling, kill switch, owner, evals, trace policy…). Low here means undeclared, not unsafe.

Run --full for the per-finding breakdown, or scaffold a ready-to-commit governance config from the scan:

release-gate audit . --emit-config -o governance.yaml
# Fill in the TODO lines, then gate every deploy:
release-gate score governance.yaml

What is release-gate?

release-gate sits between your tests and your deployment. It scans your agent code for the failure modes that only exist once an LLM is in the loop, runs evals, validates execution traces, checks cost budgets — then gives you two honest scores and one decision: PROMOTE / HOLD / BLOCK.

SonarQube checks your code. release-gate checks whether your agent change meets its release policy. They're complementary — keep your SAST suite; release-gate covers the agent layer it was never built to see (prompt-injection surfaces, cost-runaway loops, missing kill switches).

$ release-gate score governance.yaml --evals evals.yaml

  release-gate  |  Readiness Scorer  v0.9.4

  Project          customer-support-agent  v1.0.0
  Checks run       5  (5 pass, 0 warn, 0 fail)
  Evals run        7  (7 pass, 0 fail)  pass rate 100%
  Traces checked   1  (0 violations)

  Score            94 / 100   confidence: high

  Dimension Breakdown:
    safety          100  ██████████  (wt 30%)
    cost             90  █████████░  (wt 20%)
    access_control  100  ██████████  (wt 20%)
    fallback        100  ██████████  (wt 15%)
    eval_quality     85  ████████░░  (wt 10%)
    observability    80  ████████░░  (wt 5%)

  Critical failures  none

  Decision:  ✓  PROMOTE  (score 94/100)  exit 0

---

What it detects — the agent-safety rule catalog

Every finding carries a stable, citable rule id (RG-EXEC-001) that never changes when we reword a title, a one-line rationale, and a mapping to the frameworks you already answer to (OWASP LLM Top 10, NIST AI RMF). "Why did this block my release?" resolves to a rule, not a code dive. Full catalog with fixes and compliance tags: docs/RULES.md.

Two disciplines run through every rule:

• Precision over recall — we don't cry wolf. When the analyzer can't prove a real risk it stays quiet. Zero false positives on the current labeled benchmark and framework dogfood set — the 79-case corpus now carries ≥2 vulnerable and ≥2 clean look-alikes for every rule (including the v0.9.0 catalog), so that result is reproducible per rule (run python benchmark/run.py), not just asserted; and the engine stayed correctly silent across a framework dogfood (llama_index, crewAI, langgraph, open-interpreter).

• Three evidence tiers — a HIGH is watertight or it isn't a HIGH. The tier is decided by provenance we can point at, never by how a variable is spelled. CI can gate on confirmed-only.

  Tier	Max severity	What it requires
  confirmed	high	A traced origin visible in the file, with a citable chain
  inferred	medium	Real dangerous structure, origin guessed from a name — "confirm the source"
  heuristic	low	Pattern present in agent code; no flow established

  Every HIGH carries the chain a reviewer can open and check — request.json (L7) -> payload -> os.system() (L8) — and the benchmark machine-checks that every HIGH in the corpus is confirmed and provenance-backed (HIGH-tier integrity violations: 0). This came out of a real failure: a variable named payload used to be enough to assert confirmed RCE, and in AutoGPT that variable was its own HMAC-signed cache. One bad HIGH costs more trust than ten missed MEDIUMs, so a name can no longer produce one.

• How far taint reaches, and where it stops. A value is followed across a local function call, across module boundaries (a whole-program summary index resolves from x import f, and the chain names the defining file), and through the filesystem — model output written to a script that is later executed. Labeled corpus: 100% precision, 100% recall.

  It also resolves methods on classes (ai.start(...) → AI.start → AI.next → self.llm.invoke, transitively and across modules), clients held on attributes (self.llm = ChatOpenAI(...)), and container mutation (messages.append(reply)).

  Where it still stops, measured: a client built by a factory method, and model output marshalled through project-specific container/store classes — which is why gpt-engineer (a ten-hop chain) stays silent. Across the 20-repo deployed-agent corpus the taint rules produce 0 confirmed HIGHs (72 findings, all medium/low); the rule that does fire there is blast radius (RG-GATE-001) — irreversible tools with no code-level gate. Following the rest needs whole-program type inference, which we'd rather disclose than fake. We publish the numbers that didn't move, not just the ones that did.

Rule	Detects	Severity
RG-EXEC-001	Model/user output → eval/exec/os.system/a shell/subprocess — the CVE-2025-51472 RCE class	HIGH
RG-EXEC-002	pickle/marshal/dill deserialization of unverified data	MED
RG-EXEC-003	A dynamic exec/shell call in agent code, reachability unproven	LOW
RG-PROMPT-001	Untrusted text interpolated into a system prompt (OWASP LLM01)	HIGH
RG-PROMPT-002	Indirect prompt injection — retrieval/RAG, an HTTP body, or a tool return reaching the system/instruction channel, keyed on real provenance	HIGH
RG-ACTION-002	SSRF / egress — a model-controlled URL into an HTTP client	HIGH
RG-ACTION-003	Filesystem write/delete from model output (irreversible)	HIGH
RG-ACTION-004	SQL built by interpolating model output (agent-driven SQLi)	HIGH
RG-SECRET-001	Hardcoded secret / API key in source	HIGH
RG-SECRET-002	Secret/PII → prompt → third-party model — an agent-aware data-egress path (a secret in prompt content, distinct from a key used as auth) that conventional SAST often lacks the context to model	HIGH
RG-COST-001/002	LLM call / param dict with no max_tokens ceiling	LOW
RG-LOOP-001	Unbounded loop around an LLM call — the AutoGPT runaway	HIGH
RG-PARSE-001	Unvalidated model-output parse (json.loads with no try/except) — reliability	LOW
RG-TOOL-001	An agent tool's irreversible blast radius is undeclared	LOW
RG-GATE-001	An irreversible tool action with no confirmation / dry-run / human-in-loop gate	MED

Why a SAST tool can't do this: SonarQube sees eval(x) and asks "is x tainted by SQL/HTTP?" — it has no concept of "x is the model's reply." That blind spot is the entire agent layer: eval/pickle of model output, a retrieved document reaching the system role, a model-chosen URL or SQL query, a secret leaking into a prompt. Guardrails filter one input; evaluators score one output; neither blocks a release. release-gate is the gate.

See a rule fire — the reproducible demo

examples/demo-code-risk/ is a runnable before/after: a data-analysis agent that does expr = resp.choices[0].message.content; eval(expr, {"df": df}) vs. an allowlisted- aggregation fix.

pip install release-gate
git clone https://github.com/VamsiSudhakaran1/release-gate && cd release-gate
./examples/demo-code-risk/build_demo.sh            # print the demo
./examples/demo-code-risk/build_demo.sh --check    # …and assert every claim below

It builds a throwaway git repo and runs the real gate on the PR:

🔴 release-gate — AI-change review: BLOCK

Agent Code Safety: 100 → 76 (▼ -24)

Introduced by this change (not pre-existing):
  HIGH (high · confirmed)  Dangerous execution sink   agent.py:25
    ↳ eval() executes `expr`, which we traced to the model's own output at line 17.

Then it scans the same service to show both tiers side by side — the point of the product:

  • HIGH    high confidence · confirmed   Dangerous execution sink        agent.py:25
     Evidence: client.chat.completions.create() (L17) -> `expr` -> eval() (L25)

  • MEDIUM  medium confidence · inferred  Deserialization of unverified data   cache.py:26
     Evidence: `payload` -> pickle.loads() (L26) — origin unknown (name suggests external input)

Same danger shape, two different claims. The HIGH names the line the value came from, so you can open it and check us. cache.py's pickle.loads(payload) is where a name-matching scanner asserts confirmed RCE — we won't, because payload is a bare parameter and a name is not evidence (in AutoGPT that same variable held the cache's own HMAC-signed bytes).

No mockup: --check asserts every line quoted here, and CI runs it on every push, so the published demo can't drift from the engine. Live walkthrough: release-gate.com/demo.html.

---

Stop debugging AI-generated code you can't trust — release-gate pr

AI writes the diff in seconds. Then a human burns an hour deciding whether to trust it — reading every changed file, hunting for the one dangerous line, wondering if a prompt or model changed under the hood. That verification tax is where the productivity goes. Token usage is at an all-time high; shipping velocity isn't keeping up, because reviewing and trusting generated code is now the bottleneck, not writing it.

release-gate pr pays that tax down. It runs in CI on the pull request and answers one question, from evidence, not vibes: what did this change introduce that a reviewer would otherwise have to find by hand?

release-gate pr --base origin/main            # in CI, on the PR branch
release-gate pr --base origin/main --comment  # GitHub-ready markdown comment
release-gate pr --base origin/main --json     # machine output for a bot

🔴 release-gate — AI-change review: BLOCK
this change made things net-worse — see reasons

Agent Code Safety: 100 → 88 (▼ -12)

Introduced by this change (not pre-existing):
- ⚠ HIGH (high · confirmed): Dangerous execution sink   src/agent/tools.py:88
  ↳ eval() executes `reply` — the model's own output.
- ⚠ prompt changed `prompts/system.txt` — release-gate.lock not updated

Context (advisory, not blocking):
- 11 source file(s) changed, 0 test files touched
- agent code-safety -12

Inherited debt ignored (not this change's fault): 4 finding(s).

exit 1   (0 PROMOTE · 10 HOLD · 1 BLOCK)

Why you can trust this gate — and not have to debug it

This is a security tool; it's held to the standard it audits. Four properties make the verdict trustworthy on its own:

1. Every line is a fact derived from your diff — never a prediction. There is no "debug-debt score" guessing from file counts. It reports what is: this file now reaches eval() with model output; this prompt changed without a lockfile update. Facts don't cry wolf.
2. It blocks only on net-new regressions, never inherited debt. Pre-existing findings from the base branch are shown as ignored — a PR is judged on what it changed. A gate that nags about old debt gets muted, and a muted gate helps no one.
3. It's precision-calibrated. The static engine is AST + taint (not grep): it flags a sink only when model/user input can actually reach it, and grades severity by proof (confirmed vs inferred). We validated it against 18 popular agent frameworks and spent as much effort killing false positives as finding bugs — because one bad flag is how a scanner loses your trust.
4. It sees what a human diff-review structurally can't. A model or prompt change has no code fingerprint. The lockfile (AIBOM) drift check surfaces "the behavior changed but nothing in the diff shows it" — the exact class of silent change that causes 2am incidents.

What it does NOT do: it is not an AI reviewer, debugger, or fixer, and it does not add 40 inline comments. It gives one decision and the short list of things worth your attention. It's release discipline, not more noise.

Drop it into GitHub Actions — either the raw CLI:

- uses: actions/checkout@v4
  with: { fetch-depth: 0 }        # full history so the diff can be scoped
- run: pip install release-gate
- run: release-gate pr --base origin/${{ github.base_ref }} --comment >> $GITHUB_STEP_SUMMARY

…or the published Action, which also posts a sticky PR comment:

- uses: actions/checkout@v4
  with: { fetch-depth: 0 }
- uses: VamsiSudhakaran1/release-gate@v0.9.4
  with:
    command: pr
    base: origin/${{ github.base_ref }}
    pr-comment: true            # create/update one sticky comment on the PR

---

Full command & feature reference

The sections above are the whole story a new visitor needs: the problem, one command, one real finding, one GitHub Action, and how it's different. Everything else — the complete command reference, MCP server, AIBOM/drift gate, loop verification, live agent scoring, governance checks, CI recipes, evidence packs, the impact simulator, cryptographic governance, and supported model profiles — lives in docs/REFERENCE.md, so this page stays a landing page, not a manual.

---

Development

git clone https://github.com/VamsiSudhakaran1/release-gate
cd release-gate
pip install -e ".[dev]"
pytest tests/

594 tests · all passing.

---

Contributing

Found a bug? Have a feature request? Open an issue.

---

License

MIT — See LICENSE

---

Contact: [e-mail] · GitHub · Website