Live survey304 models · 14 providers · 23,592 auditsAudited implement · SAFE →What changedHow we verifyllms.txt

Agent skill · Audited from source

update-skills-from-changelog

skill · elevenlabs/skillsMIT licenceAudited 27 Oct 2026 · source 45579dc9636a

★ 511pushed 3 days ago
✓SAFEGrade B · 89/100 · 27 Oct 2026 · 45579dc
511repo stars · whole elevenlabs/skills

Is it safe? Yes — SAFE, grade B (89/100). We read this skill's source at commit 45579dc9636a and found no filesystem, network or shell access and no hard-coded secrets. What it does: Update ElevenLabs agent skills from a merged weekly changelog in elevenlabs-dx, then open a pull request in elevenlabs/skills. Trigger after a changelog merges to main on elevenlabs-dx, or when asked to update skills…

Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
More
Cite

Cite this page

ModelsAtlas. “update-skills-from-changelog (elevenlabs/skills)”. Retrieved 27 Oct 2026. https://modelsatlas.com/skills/elevenlabs/update-skills-from-changelog

Trust audit

Audit 0.4.1 · source read 27 Oct 2026

Trust score89/100✓SAFE B
0 critical0 high1 medium0 low0 info

SAFE means nothing in the source contradicts what the skill says it does. Grade B is shared by 17,406 of 18,389 audited skills.

What the source does

  • FilesystemNone observedNo reads or writes in the source
  • NetworkNone observedNo outbound calls in the source
  • ShellNone observedNo commands run in the source
  • DependenciesPinnedNothing floats
  • SecretsNone foundNo keys in source

Five audit layers

LayerWhat we checkResultFindings
L0 ProvenanceRepo owner, licence, archive status, activityWARN1
L1 Static sourceSecrets, path traversal, supply chain, hidden codeNA0
L2 InstructionsPrompt injection, authority framing, hidden directivesPASS0
L3 InventoryThe files and declarations match the sourcePASS0
L4 SandboxRun in isolation and watch behaviourSKIPPED why0

Gates applied: no behavioural pass.

Findings (1)

Audit 0.4.1 · source sha 45579dc9636a · 27 Oct 2026Report an issue or request a re-scan

Provenance

Who published it, and what the verdict is pinned to

  • Publisherelevenlabs GitHub organizationOwner of the repository this skill is read from
  • Source folderelevenlabs/skills/.agents/skills/update-skills-from-changelogThe exact folder we audited
  • LicenceMITAs the repository states it
  • Commit pinned for this verdict45579dc9636aA new audit runs when the source changes
  • Repository statusNot archivedLast push 3 days ago
  • Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

What it tells the agent

Full source on GitHub ↗

Its description: Update ElevenLabs agent skills from a merged weekly changelog in elevenlabs-dx, then open a pull request in elevenlabs/skills. Trigger after a changelog merges to main on elevenlabs-dx, or when asked to update skills…

---
name: update-skills-from-changelog
description: >
  Update ElevenLabs agent skills from a merged weekly changelog in
  elevenlabs-dx, then open a pull request in elevenlabs/skills. Trigger
  after a changelog merges to main on elevenlabs-dx, or when asked to
  update skills from changelog YYYY-MM-DD.
---

You update ElevenLabs agent skills based on a merged weekly changelog in [elevenlabs-dx](https://github.com/elevenlabs/elevenlabs-dx). Do not read GitHub issues. Open a pull request in this repository when skill files change.

Skill files are evergreen source-of-truth documentation for current behavior. Use the changelog to discover what changed, but write final `SKILL.md` and `references/*.md` content as timeless present-tense documentation.

Skill files are high-level, task-oriented guidance for working with ElevenLabs. They are not meant to mirror every nuance from the changelog or API reference. Prefer documenting primary workflows, core capabilities, and important configuration surfaces. Usually omit edge cases, precedence chains, persistence mechanics, fallback order, implementation details, and narrow exceptions unless leaving them out would make the skill materially misleading or unusable.

## Workflow

1. Resolve `CHANGELOG_DATE` from the automation trigger or user message.
2. Fetch the merged changelog from `elevenlabs-dx` `main`.
3. Apply the relevance filter. Stop if no skills are affected.
4. Read every affected skill file and reference file.
5. Verify each candidate update against canonical source documentation.
6. Apply only targeted, evergreen skill edits that pass the fit gate.
7. Self-check all edits.
8. Create `skills-update/YYYY-MM-DD`, commit, push, and open a PR.

Do not modify skills not implicated by the changelog.

## Step 1: Fetch changelog

Determine `CHANGELOG_DATE` (`YYYY-MM-DD`) from the automation trigger or user message.

Fetch the merged changelog file:

```bash
gh api "repos/elevenlabs/elevenlabs-dx/contents/fern/docs/pages/changelog/${CHANGELOG_DATE}.md?ref=main" \
  --jq '.content' | base64 -d > "/tmp/changelog-${CHANGELOG_DATE}.md"
```

If the file does not exist, stop and report that no merged changelog was found for that date.

Also read the published page when helpful:

`https://elevenlabs.io/docs/changelog#${CHANGELOG_DATE}T00:00:00.000Z`

## Step 2: Apply relevance filter

Read `/tmp/changelog-${CHANGELOG_DATE}.md` and map changes against these skills:

| Skill | What triggers an update |
| --- | --- |
| `text-to-speech` | New/deprecated models, new TTS parameters, voice settings changes, output format changes, SDK method signature changes for `text_to_speech.convert()` |
| `speech-to-text` | New transcription models, new parameters, changed response schemas, SDK method changes |
| `agents` | New LLM providers/models, new tool types, new agent config fields, conversation config schema changes, new CLI commands, widget changes, procedure endpoint/field/type changes, structured procedure step types and validation rules, draft-save or publish compilation and validation behavior, SDK method changes under `conversational_ai.agents.procedures` or `conversationalAi.agents.procedures` |
| `sound-effects` | New generation parameters, model changes, SDK method changes |
| `music` | New endpoints, new parameters, model changes |
| `voice-isolator` | New parameters, model changes, SDK method changes for `audio_isolation.convert()` |
| `realtime` | Realtime model IDs, `realtime_model` agent config, managed settings, Realtime availability by region or compliance mode, Realtime API (`/v1/convai/realtime`) or `client_secrets` changes, SDK method changes for `conversational_ai.realtime` or `conversationalAi.realtime` |
| `speech-engine` | Speech Engine WebSocket API changes, conversation token changes, SDK method changes for real-time voice conversations |
| `voice-changer` | New speech-to-speech parameters, model changes, SDK method changes for `speech_to_speech.convert()` or `speechToSpeech.convert()` |
| `setup-api-key` | Authentication flow changes, API key dashboard changes, environment variable guidance |
| `onboarding` | `elevenlabs onboard` CLI changes, approval page or sign-up flow changes, CLI install channels, SDK package names |

A change is relevant if it affects model tables, code examples, parameter documentation, configuration tables, or CLI commands documented in skills.

A change is not relevant if it only affects internal/admin APIs, optional fields with no usage-level impact, backward-compatible renames, or pricing/dashboard UI unrelated to the API key setup flow.

If no skills are affected, stop successfully without opening a pull request. Report `No skills-relevant changes for CHANGELOG_DATE`.

For each relevant item, note the affected skill and affected area, such as model table, code examples, LLM provider table, tools section, CLI section, parameter documentation, or configuration table.

## Step 3: Read current skill files

For each affected skill, read:

- `{skill}/SKILL.md`
- All files in `{skill}/references/`

Skill directories:

- `text-to-speech/` (`SKILL.md` plus `references/installation.md`, `references/streaming.md`, `references/voice-settings.md`)
- `speech-to-text/` (`SKILL.md` plus `references/installation.md`, `references/transcription-options.md`, `references/realtime-server-side.md`, `references/realtime-client-side.md`, `references/realtime-commit-strategies.md`, `references/realtime-events.md`)
- `agents/` (`SKILL.md` plus `references/installation.md`, `references/agent-configuration.md`, `references/client-tools.md`, `references/widget-embedding.md`, `references/outbound-calls.md`, `references/using-procedure-api.md`, `references/writing-procedures.md`)
- `sound-effects/` (`SKILL.md` plus `references/installation.md`)
- `music/` (`SKILL.md` plus `references/installation.md`, `references/api_reference.md`)
- `voice-isolator/` (`SKILL.md` plus `references/installation.md`)
- `realtime/` (`SKILL.md` plus `references/se

Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

Install

Pinned to the audited commit 45579dc

Download folder ↗
git clone https://github.com/elevenlabs/skills
git -C skills checkout 45579dc9636a

The author documents no install command, so this uses the repository directly. Our verdict covers commit 45579dc9636a only.

Works in

The author has not said which agents this skill supports, and we have not tested it in any.

Audit history

Every audit, every source change

  1. ✓SAFE B 89 / 100

    first audit · source 45579dc9636a

Questions about update-skills-from-changelog

What does the update-skills-from-changelog skill do?
Collections of skills for building with ElevenLabs
Is update-skills-from-changelog safe to install?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean skill reads B.
What can update-skills-from-changelog access on my machine?
The audit observed no filesystem, network or shell use at all in its source.
How current is this page?
The grade is for one exact copy of the source (45579dc9636a), read on 2026-10-27. The repository is watched, and a new audit runs when it changes — this is the first audit.

Sources

Spotted something wrong? Report a correction · Methodology

Keep going in your AI

Opens with this page as the source. We add nothing else to the prompt.