cogmeta-talon 0.8.0

Context intelligence for Agentic Development — MCP server + setup CLI

Talon

Context intelligence for Agentic Development — graph-based code context for AI coding agents.

Supported Languages

Language	Extensions	Parser Features
Python	.py	Classes, decorators, raise/except, async
TypeScript	.ts, .tsx	Classes, interfaces, arrow functions, ES imports
JavaScript	.js, .jsx, .mjs, .cjs	Arrow functions, ES imports/exports, method chains, error paths
Go	.go	Structs, receiver methods, interfaces, panic/recover
Java	.java	Annotations, constructors, throws, inheritance, enums
Rust	.rs	Structs, traits, impl blocks, enums, panic!/? error paths
C#	.cs	Classes, interfaces, attributes, constructors, throw/catch
Kotlin	.kt, .kts	Data/sealed classes, companion objects, suspend functions
Ruby	.rb	Classes, modules, include/extend, singleton methods, rescue
PHP	.php	Classes, interfaces, traits, enums, namespace imports

All languages get full-depth parsing: classes/structs, functions/methods, imports, call edges, error flow edges, method chain resolution, and cross-module call edges.

Install

One-line installer (recommended — sets up the venv, PATH links, Claude Code hooks and statusline):

curl -fsSL https://api.cogmeta.ai/install.sh | sh

Or install the wheel directly (the package is cogmeta-talon — not talon, which is an unrelated PyPI project):

pipx install cogmeta-talon

Quick start

# Authenticate with cogmeta.ai (opens browser, no copy/paste)
talon auth login

# Generate .mcp.json for your project
talon init

# Index your project into the cloud code graph
talon index

Then restart Claude Code — Talon is active. Verify with talon check (or talon doctor for a full diagnosis).

After your first run

Once you've dispatched a job, your trust dashboard surfaces every piece of data Cogmeta holds about you — self-serve, no support ticket required:

• app.cogmeta.ai/dispatch/trust.html — PII findings in your recent jobs (heuristic scanner; you confirm each redaction).
• app.cogmeta.ai/dispatch/redaction.html — every redaction you've requested + the executed_at timestamp once the TombstoneSweeper actually erases the content.
• app.cogmeta.ai/dispatch/retention.html — per-category TTL settings (jobs, conversations, messages). Data older than the TTL is auto-erased on the RetentionSweeper's hourly tick.
• app.cogmeta.ai/dispatch/jobs.html — your dispatched jobs + cost. Banners surface jobs that will expire within 3 days so you can export before they vanish.
• app.cogmeta.ai/dispatch/facts.html — Mechanism A memory facts. Every job you complete extracts up to 5 durable claims about your codebase (file paths, symbol names, error patterns, design decisions); future jobs in the same area get those facts injected into the agent's system prompt. View what's been learned, delete any of it; immediate erasure, no operator review.
• app.cogmeta.ai/dispatch/facts-considered.html — what the ranker considered but didn't inject. Promote one to boost its confidence so it crosses the T1 floor next time; suppress one to delete it entirely. Every click tunes your ranker — the precision-feedback loop that compounds over time.
• app.cogmeta.ai/audit-log.html — hash-chained audit trail (tamper-evident; verify locally with the chain's prev_hash links).

For machine-readable GDPR Art 15 export: POST /cloud/v1/dispatch/user/data/export returns the bundle inline. Schema is published at app.cogmeta.ai/dispatch/export-bundle/v1.0/ — pin schema_version and DPO automation evolves additively.