Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
55 changes: 55 additions & 0 deletions AGENTS.md
Original file line number Diff line number Diff line change
@@ -1,5 +1,60 @@
## AgenticLens Development Reference

## Ecosystem Context

### Role in DeepAgentLabs

`agenticlens` is the observability, evaluation, and operational intelligence
layer in DeepAgentLabs. It helps teams understand what ran, what the model and
agent did, what context and tools were involved, what it cost, and whether the
system performed well over time.

### Owns

- Runtime instrumentation, trace capture, analysis, evaluation, reporting, and
evidence-backed recommendations
- The flagship Python reference implementation for emitting and working with AI
Operations Specification-compatible artifacts
- Export surfaces and analysis workflows that turn raw traces into actionable
operational findings

### Does Not Own

- The normative definition of the shared contract — that belongs in
`ai-operations-spec`
- Fault injection or resilience experimentation — that belongs in
`agentic-chaos`
- Pre-action agent supervision and decision interception — that belongs in
`agentic-sidecar`
- A generic MCP server or remote orchestration surface — that belongs in
`deep-agentic-core-mcp`

### Integrates With

- `ai-operations-spec` as the canonical object model and long-term
interoperability target
- `agentic-chaos` for ingesting and analyzing resilience/failure evidence
- `agentic-sidecar` when decision and escalation events need to become part of
the operational record
- `deep-agentic-core-mcp` as a thin delivery surface for AgenticLens-powered
workflows and analysis tools

### Current Roadmap Focus

The next milestone centers on judge calibration, evaluation dataset management,
and experiments/statistical comparison. New work here should deepen evidence
quality and repeatable analysis, not drift into spec stewardship or chaos
simulation.

### Before You Build Here

- If a change defines a cross-package object, event meaning, or schema, move it
to `ai-operations-spec` instead of making AgenticLens the de facto standard
- Prefer consuming chaos, sidecar, or MCP artifacts through clean boundaries
rather than reimplementing sibling-package behavior here
- Keep findings evidence-backed: this package should explain and evaluate what
happened, not become a policy engine or fault injector

## Build and Run

- Install: `make install` (runs `uv sync --extra dev --extra docs`)
Expand Down
Loading