CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

cicd-secrets-exfil

Extracting CI secrets / OIDC tokens once you have code execution in a build job — echo/printenv exfil, log-masking bypass (base64, char-split, reversal), OIDC token abuse to assume cloud roles, GITHUB_TOKEN / CI_JOB_TOKEN scope abuse, cache / artifact secret leakage, provenance pivot.

Quality

79%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./packages/decepticon/decepticon/skills/standard/exploit/cicd/cicd-secrets-exfil/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

71%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.

A high-density, highly actionable body that assumes competence and provides copy-paste-ready commands across GitHub, GitLab, and CircleCI, with an explicit safety decision gate. The main weaknesses are the monolithic single-file structure (no progressive disclosure despite 185 lines, plus dangling references to sibling skills) and a few placeholder-stubbed paths (GCP exchange).

Suggestions

Split provider-specific detail into reference files (e.g., references/gitlab-job-token.md, references/detection-signatures.md) and link them from the body, turning the 185-line monolith into an overview with one-level-deep references.

Complete or remove stubbed guidance: either give the GCP Workload Identity Federation exchange commands or drop the section, and resolve placeholders like <COLLAB>, <OTHER_PROJECT_ID>, and $AWS_ROLE_ARN with sourcing instructions.

Move time-sensitive claims ("default historically", CVE-2025-30066, the tj-actions incident) into a clearly-labeled incidents/version-notes section, and fix or remove the dangling references to poisoned-pipeline-execution/ and github-actions-injection/.

DimensionReasoningScore

Conciseness

The body is dense and command-first, assuming Claude's competence (env-var maps are bare comments, no library tutorials). A few things could be trimmed: the SLSA narrative sentence, re-explanations like "GITHUB_TOKEN is short-lived (job duration) and not exfilable to durable infra", and time-sensitive claims ("default historically", CVE-2025-30066) that are not quarantined in a deprecated/old-patterns section.

4 / 5

Actionability

Nearly every section gives copy-paste-ready commands with real flags (e.g., the curl OIDC mint, "aws sts assume-role-with-web-identity --duration-seconds 900", the gh API calls). Minor gaps keep it below 5: the GCP Workload Identity Federation exchange is only a comment, and $AWS_ROLE_ARN, <COLLAB>, <OTHER_PROJECT_ID>, and RUN_ID are unresolved placeholders.

4 / 5

Workflow Clarity

Sections form a coherent progression (surface map → env inventory → masking bypass → OIDC → token abuse → cleanup) and the numbered Decision gate supplies explicit validation checkpoints (proof-not-secret limits, scrub artifacts before declaring done), satisfying the destructive-operation validation requirement. It is not 5 because the main exploitation flow is never explicitly ordered — sequence must be inferred from heading arrangement.

4 / 5

Progressive Disclosure

Headers are clear, but this is a single ~185-line file with no bundle files at all: the per-provider deep-dives (GitLab job-token pivot, CircleCI), the detection-signatures table, and the tools table are natural candidates for references/. Additionally, the body cites "poisoned-pipeline-execution/" and "github-actions-injection/" which do not exist in the bundle, and the >50-line simple-skill exception does not apply.

3 / 5

Total

15

/

20

Passed

Description

78%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.

A strong description: comprehensive and concrete about what the skill does, with a well-defined niche and good natural trigger terms. Its main weaknesses are the second-person phrasing in the precondition clause and the absence of explicit user-intent trigger phrases.

DimensionReasoningScore

Specificity

The description lists multiple concrete, comprehensive actions ("echo/printenv exfil, log-masking bypass (base64, char-split, reversal), OIDC token abuse to assume cloud roles, GITHUB_TOKEN / CI_JOB_TOKEN scope abuse, cache / artifact secret leakage, provenance pivot"), which matches the 5 anchor. It is reduced by 1 per the judging guidelines because "once you have code execution in a build job" uses second-person voice instead of third person.

4 / 5

Completeness

"What" is explicitly and comprehensively answered, and "when" is present as an explicit precondition ("once you have code execution in a build job") rather than merely implied. It is not a 5 because the when-guidance is a state condition rather than concrete user-intent trigger phrases ("use when the user mentions...").

4 / 5

Trigger Term Quality

Good keyword coverage with natural terms a user would say: "ci secrets", "oidc", "printenv", "masking bypass", "github_token", "ci_job_token", "cache", "artifact". Missing common variations/synonyms such as "github actions", "gitlab", "runner", "exfiltrate", so it falls between the 4 and 5 anchors, closer to 4.

4 / 5

Distinctiveness Conflict Risk

Clear niche with distinct triggers — OIDC token exchange, CI_JOB_TOKEN, log-masking bypass, provenance pivot are highly specific tokens unlikely to fire for any other skill. Minimal conflict risk, matching the 5 anchor.

5 / 5

Total

17

/

20

Passed

Validation

93%

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

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

Total

15

/

16

Passed

Repository

PurpleAILAB/Decepticon

Path

packages/decepticon/decepticon/skills/standard/exploit/cicd/cicd-secrets-exfil/SKILL.md

Reviewed

Table of Contents

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.