Skip to main content

ai-hats-observe

A standalone session-logging engine: open a session on a bare directory, append trace entries, write an incremental audit, and finalize a versioned metrics.json — with no configuration, no worktree engine, and no ai-hats.yaml.

ai-hats-observe is the observability core extracted from the ai-hats framework (ADR-0014 Phase 1, T15). It has no dependency on the ai-hats integrator: everything below runs against a plain directory. Its only runtime dependency is ai-hats-core (dependency-free filesystem primitives + the migration seam).

What it owns

  • Session lifecycleSessionManager / Session: allocate a session directory, resolve its artifact paths, list/filter past sessions.
  • Trace + audit writerlog_trace, init_audit, append_audit, finalize_audit; the enriched AuditWriter that turns a transcript into audit.md + a versioned metrics.json.
  • Versioned trace/audit schema — a schema_version on the metrics surface, with a wired (initially empty) migration seam.
  • Surface-agnostic parsing — a TranscriptParser adapter: the AuditWriter holds no provider-specific parsing; the concrete ClaudeParser (structured JSONL + trace-chrome fallback) is one implementation. A new surface (Gemini, or a future CLI) plugs in its own parser — the writer never changes.

Session privacy

New session directories are created with mode 0700. Sensitive artifacts are created and rewritten with mode 0600, including prompts, role materialization, traces, audits, metrics, transcripts, reasoning, usage, and raw PTY captures. These permissions are enforced independently of the caller's umask and apply to every provider that uses the observe session API. Existing session trees are not migrated or rewritten.

Extending to a new surface

The parser is carried by the session's provider, not a central registry: a Provider yields its TranscriptParser, and the integrator injects it into the AuditWriter. Adding a surface means shipping a parser that satisfies the TranscriptParser protocol and a provider that names it — no edit to ai-hats-observe internals.

Dependency direction

ai-hats-observe imports only ai-hats-core and the standard library. It never imports the ai-hats integrator; ai-hats imports from here. The boundary is enforced by an AST import-lint (test_observe_boundary.py).

Release files for ai-hats-observe 0.10.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for ai-hats-observe 0.10.0
File Size Uploaded
ai_hats_observe-0.10.0.tar.gz 99.1 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for ai-hats-observe 0.10.0
File Interpreter ABI Platform
ai_hats_observe-0.10.0-py3-none-any.whl Python 3 none any Details

Total release size: 171.6 kB

Release files / ai_hats_observe-0.10.0.tar.gz

Download URL ai_hats_observe-0.10.0.tar.gz
Size 99.1 kB
Tags Source
SHA-256 checksum
How to use checksums
a7be151191191a2c742762e4f32b3a9470d8e3640271497702bb56d6e1e84d70
BLAKE2b-256 checksum
How to use checksums
143cb7c82f2cb2c235f7a757f3b2da3da1b55ecc6e22af804cae6957f8e672e8
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.

Transparency log

Release files / ai_hats_observe-0.10.0-py3-none-any.whl

Download URL ai_hats_observe-0.10.0-py3-none-any.whl
Size 72.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
f5eef3832c932a4626e8ce57b3c0a4cbebd5f3259ed036c5795d4a0f87115031
BLAKE2b-256 checksum
How to use checksums
3d0151a061d9afde32eb08b94b072ed852c4e5b5a10169180ff3f808893770f1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.

Transparency log

Release history Release notifications | RSS feed

0.14.0

2 release files

0.12.0

2 release files

This release

0.10.0 This release

2 release files

0.9.0

2 release files

0.8.0

2 release files

0.7.0

2 release files

0.6.1

2 release files

0.6.0

2 release files

0.5.0

2 release files

0.4.0

2 release files

0.3.0

2 release files

0.2.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page