v0.3.4

v0.3.4

Changes

@tanstack/intent

0.3.4

Patch Changes

• #190 45c863d - Improve the maintainer skill-authoring workflow:

  - Add `validate --set-version <version>`, which stamps `metadata.library_version` across matched skills via YAML surgery (preserving comments, line endings, and body) and then re-validates.
  - Correct the `generate-skill` meta-skill so it no longer emits slash-path `name` values that fail validation, and the `skill-staleness-check` meta-skill so it points at the real `intent stale` command instead of a non-existent script.
  

• #188 2a25b36 - Refuse to load a Yarn PnP runtime resolved from within node_modules; only a project root or ancestor .pnp.cjs is trusted.

Packages

• @tanstack/intent@0.3.4

v0.3.3

v0.3.3

Changes

@tanstack/intent

0.3.3

Patch Changes

• #185 b7920e9 - Fix intent list suppressing the allow-all risk warning.

  When intent.skills is set to allow-all ("*"), intent list --no-notices and INTENT_NO_NOTICES=1 no longer hide the warning that all skill sources are permitted. This banner is a security-relevant signal, not a migration tip, so it's now excluded from suppression while other notices remain suppressible as before.

• #187 fb3c08b - Fix potential hangs on slow or large environments:

  • Bound the npm registry staleness check with a request timeout so intent list and other staleness checks can't hang indefinitely on a slow or unreachable registry.
  • Bound global package-manager detection with a command timeout so it can't hang indefinitely when the environment's global node_modules is slow to resolve.
  • Avoid enumerating a workspace package's entire skill tree just to check whether it has any skills, reducing filesystem work in large monorepos.

Packages

• @tanstack/intent@0.3.3

@tanstack/intent@0.3.4

Patch Changes

• #190 45c863d - Improve the maintainer skill-authoring workflow:

  - Add `validate --set-version <version>`, which stamps `metadata.library_version` across matched skills via YAML surgery (preserving comments, line endings, and body) and then re-validates.
  - Correct the `generate-skill` meta-skill so it no longer emits slash-path `name` values that fail validation, and the `skill-staleness-check` meta-skill so it points at the real `intent stale` command instead of a non-existent script.
  

• #188 2a25b36 - Refuse to load a Yarn PnP runtime resolved from within node_modules; only a project root or ancestor .pnp.cjs is trusted.

@tanstack/intent@0.3.3

Patch Changes

• #185 b7920e9 - Fix intent list suppressing the allow-all risk warning.

  When intent.skills is set to allow-all ("*"), intent list --no-notices and INTENT_NO_NOTICES=1 no longer hide the warning that all skill sources are permitted. This banner is a security-relevant signal, not a migration tip, so it's now excluded from suppression while other notices remain suppressible as before.

• #187 fb3c08b - Fix potential hangs on slow or large environments:

  • Bound the npm registry staleness check with a request timeout so intent list and other staleness checks can't hang indefinitely on a slow or unreachable registry.
  • Bound global package-manager detection with a command timeout so it can't hang indefinitely when the environment's global node_modules is slow to resolve.
  • Avoid enumerating a workspace package's entire skill tree just to check whether it has any skills, reducing filesystem work in large monorepos.

v0.3.2

v0.3.2

Changes

@tanstack/intent

0.3.2

Patch Changes

• #180 d67a5d6 - Add proactive skill catalogs to installed Intent hooks.

  intent hooks install now installs session-start catalog hooks for supported agents alongside the existing edit gate. Agents see the allowed local Intent skills at session start, resume, clear, and compact where the agent supports those lifecycle events, then still need to run intent load before editing.

  The generated hook loads the catalog through the Intent CLI with agent audience redaction instead of importing package code from the target repository.

Packages

• @tanstack/intent@0.3.2

@tanstack/intent@0.3.2

Patch Changes

• #180 d67a5d6 - Add proactive skill catalogs to installed Intent hooks.

  intent hooks install now installs session-start catalog hooks for supported agents alongside the existing edit gate. Agents see the allowed local Intent skills at session start, resume, clear, and compact where the agent supports those lifecycle events, then still need to run intent load before editing.

  The generated hook loads the catalog through the Intent CLI with agent audience redaction instead of importing package code from the target repository.

v0.3.1

v0.3.1

Changes

@tanstack/intent

0.3.1

Patch Changes

• #176 5419b02 - Add agent-safe hidden skill source handling to intent list.

  intent list now detects agent sessions with std-env and redacts unlisted allowlist candidates from agent-facing output. Agents see allowed skills plus a count-only hidden-source notice, while intent list --show-hidden reveals hidden source names only outside agent sessions.

  Agent-mode JSON output includes hiddenSourceCount but leaves hiddenSources empty, preventing structured output from leaking package names that could be added to package.json#intent.skills.

Packages

• @tanstack/intent@0.3.1

v0.3.0

v0.3.0

Changes

@tanstack/intent

0.3.0

Minor Changes

• #174 089a355 - Add intent hooks install for supported AI coding agents.

  This adds lifecycle-hook installation for supported agents, including project/user scope handling, generated hook runner scripts, and agent-specific enforcement policy. It also documents the hook setup flow and adds eval/test coverage for hooked intent discovery.

Packages

• @tanstack/intent@0.3.0

@tanstack/intent@0.3.1

Patch Changes

• #176 5419b02 - Add agent-safe hidden skill source handling to intent list.

  intent list now detects agent sessions with std-env and redacts unlisted allowlist candidates from agent-facing output. Agents see allowed skills plus a count-only hidden-source notice, while intent list --show-hidden reveals hidden source names only outside agent sessions.

  Agent-mode JSON output includes hiddenSourceCount but leaves hiddenSources empty, preventing structured output from leaking package names that could be added to package.json#intent.skills.

@tanstack/intent@0.3.0

Minor Changes

• #174 089a355 - Add intent hooks install for supported AI coding agents.

  This adds lifecycle-hook installation for supported agents, including project/user scope handling, generated hook runner scripts, and agent-specific enforcement policy. It also documents the hook setup flow and adds eval/test coverage for hooked intent discovery.