tokenjam 0.2.0__tar.gz → 0.2.1__tar.gz

{tokenjam-0.2.0 → tokenjam-0.2.1}/.gitignore

@@ -39,8 +39,8 @@ sdk-ts/dist/
 # Secrets and env files
 .env
 .env.*
-ocw.toml
-.ocw/
+tj.toml
+.tj/
 
 # Claude Code
 .claude/

{tokenjam-0.2.0 → tokenjam-0.2.1}/CHANGELOG.md

@@ -8,7 +8,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 ## [0.1.7] - 2026-04-13
 
 ### Added
-- **MCP server (`tj mcp`)** — stdio-based Model Context Protocol server giving Claude Code direct access to OCW observability data. 13 tool handlers: status, traces, alerts, budget headroom, cost summary, drift report, tool stats, trace detail, acknowledge alerts, setup project, list sessions, open dashboard. Dual-mode operation: routes queries through REST API when `tj serve` is running, falls back to read-only DuckDB otherwise. Auto-starts `tj serve` on demand.
+- **MCP server (`tj mcp`)** — stdio-based Model Context Protocol server giving Claude Code direct access to TokenJam observability data. 13 tool handlers: status, traces, alerts, budget headroom, cost summary, drift report, tool stats, trace detail, acknowledge alerts, setup project, list sessions, open dashboard. Dual-mode operation: routes queries through REST API when `tj serve` is running, falls back to read-only DuckDB otherwise. Auto-starts `tj serve` on demand.
 - **Claude Code integration (`tj onboard --claude-code`)** — one-command setup for Claude Code telemetry. Configures OTLP log exporter in `~/.claude/settings.json`, sets project-level `OTEL_RESOURCE_ATTRIBUTES`, adds Docker-compatible endpoint to shell env, and optionally installs background daemon. Re-runs resync the auth header to fix 401s without manual setup.
 - **Logs ingestion (`POST /v1/logs`)** — new OTLP log endpoint that converts Claude Code log events (`api_request`, `tool_result`, `api_error`, `user_prompt`, `tool_decision`) into NormalizedSpans with deterministic trace/span IDs. Spans flow through the standard ingest pipeline for cost, alerts, and drift.
 - **`tj drift` CLI** — behavioral drift report with Rich table output showing baseline vs latest session Z-scores per dimension (input tokens, output tokens, duration, tool call count, tool sequence similarity). Color-coded thresholds, `--json` support, exit code 1 if drift detected.
@@ -106,7 +106,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 
 ### Added
 - `tj stop` command — graceful shutdown of daemon or background process
-- `tj uninstall` command — clean removal of all OCW data, config, and daemon
+- `tj uninstall` command — clean removal of all TokenJam data, config, and daemon
 - 16 runnable example agents across 4 tiers: single provider, single framework, multi-agent, and alerts/drift demos
 - API fallback backend (`ApiBackend`) so CLI works while `tj serve` holds the DB lock
 

{tokenjam-0.2.0 → tokenjam-0.2.1}/CLAUDE.md

@@ -127,7 +127,7 @@ Post-ingest hooks run synchronously after each span is written to DB:
 | `tj drift` | `cmd_drift.py` | Show drift baselines and Z-scores for recent sessions |
 | `tj demo [scenario]` | `cmd_demo.py` | Run Agent Incident Library scenarios (zero-config, no API keys). `tj demo` lists all; `tj demo retry-loop` runs one |
 | `tj mcp` | `cmd_mcp.py` | Start the stdio MCP server for Claude Code integration |
-| `tj uninstall` | `cmd_uninstall.py` | Remove all OCW data, config, and daemon |
+| `tj uninstall` | `cmd_uninstall.py` | Remove all TokenJam data, config, and daemon |
 | `tj doctor` | `cmd_doctor.py` | Health checks (config, DB, secrets, webhooks, drift readiness, schema-vs-capture consistency). Exit 0 = ok, 1 = warnings, 2 = errors |
 
 All commands support `--json` for machine-readable output. Commands that query alerts use exit code 1 if active (unacknowledged, unsuppressed) alerts exist.

{tokenjam-0.2.0 → tokenjam-0.2.1}/Makefile

@@ -10,9 +10,9 @@ test-e2e:
 	pytest tests/e2e/
 
 lint:
-	ruff check ocw/
+	ruff check tokenjam/
 
 typecheck:
-	mypy ocw/
+	mypy tokenjam/
 
 all: lint typecheck test

{tokenjam-0.2.0 → tokenjam-0.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tokenjam
-Version: 0.2.0
+Version: 0.2.1
 Summary: TokenJam — local-first OTel-native observability for Autonomous AI agents
 Project-URL: Homepage, https://opencla.watch
 Project-URL: Repository, https://github.com/Metabuilder-Labs/openclawwatch
@@ -56,7 +56,7 @@ Description-Content-Type: text/markdown
 
 <div align="center">
 
-<img src="https://opencla.watch/icon.svg" alt="TokenJam" width="72" height="72">
+<img src="https://tokenjam.dev/icon.svg" alt="TokenJam" width="72" height="72">
 
 # TokenJam
 
@@ -122,7 +122,7 @@ For any Python agent — Anthropic, OpenAI, Gemini, Bedrock, LangChain, CrewAI,
 ```bash
 pip install tokenjam
 tj onboard    # creates config, generates ingest secret
-ocw doctor     # verify your setup
+tj doctor     # verify your setup
 ```
 
 ```python
@@ -211,12 +211,12 @@ https://github.com/user-attachments/assets/b94d13f6-1432-40d4-b093-6958d74f0e65
 
 ```bash
 tj status           # current state, cost, active alerts
-ocw traces           # full span history with waterfall view
-ocw cost --since 7d  # cost breakdown by agent, model, day
-ocw alerts           # everything that fired while you were away
-ocw budget           # view and set daily/session cost limits
-ocw drift            # behavioral drift Z-scores vs baseline
-ocw tools            # tool call history with error rates
+tj traces           # full span history with waterfall view
+tj cost --since 7d  # cost breakdown by agent, model, day
+tj alerts           # everything that fired while you were away
+tj budget           # view and set daily/session cost limits
+tj drift            # behavioral drift Z-scores vs baseline
+tj tools            # tool call history with error rates
 tj serve            # start the web UI + REST API
 ```
 
@@ -239,7 +239,7 @@ No signup, no cloud — runs entirely on your machine.
 
 ---
 
-## ocw vs LangSmith vs Langfuse
+## tj vs LangSmith vs Langfuse
 
 LangSmith and Langfuse are excellent for tracing LLM API calls and running evals on chat outputs. `tj` solves a different problem: **autonomous agents running unsupervised with real-world consequences**.
 
@@ -307,14 +307,14 @@ The MCP server gives Claude Code direct access to your observability data inside
 | `get_tool_stats` | Tool call counts and average duration |
 | `get_drift_report` | Drift baseline vs latest session |
 | `acknowledge_alert` | Mark an alert as acknowledged |
-| `setup_project` | Configure a project for OCW telemetry |
+| `setup_project` | Configure a project for TokenJam telemetry |
 | `open_dashboard` | Open the web UI (starts `tj serve` if needed) |
 
 The MCP server opens DuckDB read-only — no lock conflicts with `tj serve`.
 
 **Per-project tagging** — after installing globally, ask Claude Code:
 
-> "Set up OCW for this project"
+> "Set up TokenJam for this project"
 
 Claude calls `setup_project`, which writes `.claude/settings.json` with the right `OTEL_RESOURCE_ATTRIBUTES` for this project.
 
@@ -330,8 +330,8 @@ tj onboard --codex
 `tj onboard --codex` is project-agnostic. It writes to `~/.codex/config.toml` (Codex's single global config), so you only run it once — not once per project. Codex hardcodes `service.name="codex_exec"` in its binary, so all sessions appear under the same agent ID regardless of which repo you're working in.
 
 `tj onboard --codex`:
-- Writes an `[otel]` block and `[mcp_servers.ocw]` to `~/.codex/config.toml`
-- Registers the MCP server so Codex can call OCW tools directly
+- Writes an `[otel]` block and `[mcp_servers.tj]` to `~/.codex/config.toml`
+- Registers the MCP server so Codex can call TokenJam tools directly
 - Installs the background daemon (launchd / systemd)
 
 **Codex must be restarted** after running `tj onboard --codex`.
@@ -345,14 +345,14 @@ The same 13 MCP tools available to Claude Code are available to Codex after rest
 ### Uninstalling
 
 ```bash
-# Remove all OCW data, config, daemon, MCP registration, and env vars:
-ocw uninstall --yes
+# Remove all TokenJam data, config, daemon, MCP registration, and env vars:
+tj uninstall --yes
 
 # Then remove the package:
 pip uninstall tokenjam -y
 ```
 
-`tj uninstall` cleans up everything set by `tj onboard --claude-code`: daemon, MCP server, `~/.ocw/`, `~/.config/ocw/`, OTLP env vars in `~/.claude/settings.json`, `OTEL_RESOURCE_ATTRIBUTES` in every onboarded project's `.claude/settings.json`, and the harness env block in `~/.zshrc`.
+`tj uninstall` cleans up everything set by `tj onboard --claude-code`: daemon, MCP server, `~/.tj/`, `~/.config/tj/`, OTLP env vars in `~/.claude/settings.json`, `OTEL_RESOURCE_ATTRIBUTES` in every onboarded project's `.claude/settings.json`, and the harness env block in `~/.zshrc`.
 
 ---
 
@@ -397,7 +397,7 @@ Full framework support guide: [docs/framework-support.md](docs/framework-support
 Configure where alerts go. Multiple channels work simultaneously.
 
 ```toml
-# .ocw/config.toml
+# .tj/config.toml
 
 [[alerts.channels]]
 type = "ntfy"
@@ -438,10 +438,10 @@ Full event table and configuration: [docs/nemoclaw-integration.md](docs/nemoclaw
 ## Export and integrate
 
 ```bash
-ocw export --format otlp       # forward to Grafana, Datadog, any OTel backend
-ocw export --format openevals  # openevals / agentevals trajectory evaluation
-ocw export --format json       # NDJSON
-ocw export --format csv
+tj export --format otlp       # forward to Grafana, Datadog, any OTel backend
+tj export --format openevals  # openevals / agentevals trajectory evaluation
+tj export --format json       # NDJSON
+tj export --format csv
 ```
 
 Prometheus metrics at `http://127.0.0.1:7391/metrics` when `tj serve` is running.
@@ -478,7 +478,7 @@ flowchart TD
     Alerts --> DB
     Schema --> DB
 
-    DB --> CLI["ocw CLI"]
+    DB --> CLI["tj CLI"]
     DB --> API["REST API + Web UI\n:7391"]
     DB --> MCP["MCP Server\n13 tools"]
     DB --> Prom["Prometheus\n:7391/metrics"]
@@ -491,7 +491,7 @@ Full architecture deep-dive — design principles, SDK internals, alert system,
 ## Configuration
 
 ```toml
-# .ocw/config.toml — generated by tj onboard
+# .tj/config.toml — generated by tj onboard
 
 [defaults.budget]
 daily_usd = 10.00
@@ -518,7 +518,7 @@ completions  = false
 tool_outputs = false
 
 [storage]
-path           = "~/.ocw/telemetry.duckdb"
+path           = "~/.tj/telemetry.duckdb"
 retention_days = 90
 ```
 
@@ -559,18 +559,18 @@ See [`examples/README.md`](examples/README.md) for the full list.
 Reproducible AI agent failures you can run in 30 seconds. No API keys, no config, no setup.
 
 ```bash
-ocw demo                     # list all scenarios
-ocw demo retry-loop          # run one
-ocw demo retry-loop --json   # machine-readable output
+tj demo                     # list all scenarios
+tj demo retry-loop          # run one
+tj demo retry-loop --json   # machine-readable output
 ```
 
-| Scenario | What goes wrong | What OCW catches |
+| Scenario | What goes wrong | What TokenJam catches |
 |---|---|---|
 | [`retry-loop`](incidents/retry-loop/README.md) | Agent retries a failing tool in a loop, burning time and tokens | `retry_loop` + `failure_rate` alerts fire automatically |
 | [`surprise-cost`](incidents/surprise-cost/README.md) | Model silently escalates from Haiku to Opus mid-chain | Per-model cost breakdown shows the $3+ you didn't expect |
 | [`hallucination-drift`](incidents/hallucination-drift/README.md) | Agent behavior shifts — different tokens, different tools | `drift_detected` alert fires with Z-scores at session end |
 
-Each scenario runs against an in-memory backend and produces a side-by-side comparison: what `print()` shows vs. what OCW reveals.
+Each scenario runs against an in-memory backend and produces a side-by-side comparison: what `print()` shows vs. what TokenJam reveals.
 
 ---
 

{tokenjam-0.2.0 → tokenjam-0.2.1}/README.md

@@ -1,6 +1,6 @@
 <div align="center">
 
-<img src="https://opencla.watch/icon.svg" alt="TokenJam" width="72" height="72">
+<img src="https://tokenjam.dev/icon.svg" alt="TokenJam" width="72" height="72">
 
 # TokenJam
 
@@ -66,7 +66,7 @@ For any Python agent — Anthropic, OpenAI, Gemini, Bedrock, LangChain, CrewAI,
 ```bash
 pip install tokenjam
 tj onboard    # creates config, generates ingest secret
-ocw doctor     # verify your setup
+tj doctor     # verify your setup
 ```
 
 ```python
@@ -155,12 +155,12 @@ https://github.com/user-attachments/assets/b94d13f6-1432-40d4-b093-6958d74f0e65
 
 ```bash
 tj status           # current state, cost, active alerts
-ocw traces           # full span history with waterfall view
-ocw cost --since 7d  # cost breakdown by agent, model, day
-ocw alerts           # everything that fired while you were away
-ocw budget           # view and set daily/session cost limits
-ocw drift            # behavioral drift Z-scores vs baseline
-ocw tools            # tool call history with error rates
+tj traces           # full span history with waterfall view
+tj cost --since 7d  # cost breakdown by agent, model, day
+tj alerts           # everything that fired while you were away
+tj budget           # view and set daily/session cost limits
+tj drift            # behavioral drift Z-scores vs baseline
+tj tools            # tool call history with error rates
 tj serve            # start the web UI + REST API
 ```
 
@@ -183,7 +183,7 @@ No signup, no cloud — runs entirely on your machine.
 
 ---
 
-## ocw vs LangSmith vs Langfuse
+## tj vs LangSmith vs Langfuse
 
 LangSmith and Langfuse are excellent for tracing LLM API calls and running evals on chat outputs. `tj` solves a different problem: **autonomous agents running unsupervised with real-world consequences**.
 
@@ -251,14 +251,14 @@ The MCP server gives Claude Code direct access to your observability data inside
 | `get_tool_stats` | Tool call counts and average duration |
 | `get_drift_report` | Drift baseline vs latest session |
 | `acknowledge_alert` | Mark an alert as acknowledged |
-| `setup_project` | Configure a project for OCW telemetry |
+| `setup_project` | Configure a project for TokenJam telemetry |
 | `open_dashboard` | Open the web UI (starts `tj serve` if needed) |
 
 The MCP server opens DuckDB read-only — no lock conflicts with `tj serve`.
 
 **Per-project tagging** — after installing globally, ask Claude Code:
 
-> "Set up OCW for this project"
+> "Set up TokenJam for this project"
 
 Claude calls `setup_project`, which writes `.claude/settings.json` with the right `OTEL_RESOURCE_ATTRIBUTES` for this project.
 
@@ -274,8 +274,8 @@ tj onboard --codex
 `tj onboard --codex` is project-agnostic. It writes to `~/.codex/config.toml` (Codex's single global config), so you only run it once — not once per project. Codex hardcodes `service.name="codex_exec"` in its binary, so all sessions appear under the same agent ID regardless of which repo you're working in.
 
 `tj onboard --codex`:
-- Writes an `[otel]` block and `[mcp_servers.ocw]` to `~/.codex/config.toml`
-- Registers the MCP server so Codex can call OCW tools directly
+- Writes an `[otel]` block and `[mcp_servers.tj]` to `~/.codex/config.toml`
+- Registers the MCP server so Codex can call TokenJam tools directly
 - Installs the background daemon (launchd / systemd)
 
 **Codex must be restarted** after running `tj onboard --codex`.
@@ -289,14 +289,14 @@ The same 13 MCP tools available to Claude Code are available to Codex after rest
 ### Uninstalling
 
 ```bash
-# Remove all OCW data, config, daemon, MCP registration, and env vars:
-ocw uninstall --yes
+# Remove all TokenJam data, config, daemon, MCP registration, and env vars:
+tj uninstall --yes
 
 # Then remove the package:
 pip uninstall tokenjam -y
 ```
 
-`tj uninstall` cleans up everything set by `tj onboard --claude-code`: daemon, MCP server, `~/.ocw/`, `~/.config/ocw/`, OTLP env vars in `~/.claude/settings.json`, `OTEL_RESOURCE_ATTRIBUTES` in every onboarded project's `.claude/settings.json`, and the harness env block in `~/.zshrc`.
+`tj uninstall` cleans up everything set by `tj onboard --claude-code`: daemon, MCP server, `~/.tj/`, `~/.config/tj/`, OTLP env vars in `~/.claude/settings.json`, `OTEL_RESOURCE_ATTRIBUTES` in every onboarded project's `.claude/settings.json`, and the harness env block in `~/.zshrc`.
 
 ---
 
@@ -341,7 +341,7 @@ Full framework support guide: [docs/framework-support.md](docs/framework-support
 Configure where alerts go. Multiple channels work simultaneously.
 
 ```toml
-# .ocw/config.toml
+# .tj/config.toml
 
 [[alerts.channels]]
 type = "ntfy"
@@ -382,10 +382,10 @@ Full event table and configuration: [docs/nemoclaw-integration.md](docs/nemoclaw
 ## Export and integrate
 
 ```bash
-ocw export --format otlp       # forward to Grafana, Datadog, any OTel backend
-ocw export --format openevals  # openevals / agentevals trajectory evaluation
-ocw export --format json       # NDJSON
-ocw export --format csv
+tj export --format otlp       # forward to Grafana, Datadog, any OTel backend
+tj export --format openevals  # openevals / agentevals trajectory evaluation
+tj export --format json       # NDJSON
+tj export --format csv
 ```
 
 Prometheus metrics at `http://127.0.0.1:7391/metrics` when `tj serve` is running.
@@ -422,7 +422,7 @@ flowchart TD
     Alerts --> DB
     Schema --> DB
 
-    DB --> CLI["ocw CLI"]
+    DB --> CLI["tj CLI"]
     DB --> API["REST API + Web UI\n:7391"]
     DB --> MCP["MCP Server\n13 tools"]
     DB --> Prom["Prometheus\n:7391/metrics"]
@@ -435,7 +435,7 @@ Full architecture deep-dive — design principles, SDK internals, alert system,
 ## Configuration
 
 ```toml
-# .ocw/config.toml — generated by tj onboard
+# .tj/config.toml — generated by tj onboard
 
 [defaults.budget]
 daily_usd = 10.00
@@ -462,7 +462,7 @@ completions  = false
 tool_outputs = false
 
 [storage]
-path           = "~/.ocw/telemetry.duckdb"
+path           = "~/.tj/telemetry.duckdb"
 retention_days = 90
 ```
 
@@ -503,18 +503,18 @@ See [`examples/README.md`](examples/README.md) for the full list.
 Reproducible AI agent failures you can run in 30 seconds. No API keys, no config, no setup.
 
 ```bash
-ocw demo                     # list all scenarios
-ocw demo retry-loop          # run one
-ocw demo retry-loop --json   # machine-readable output
+tj demo                     # list all scenarios
+tj demo retry-loop          # run one
+tj demo retry-loop --json   # machine-readable output
 ```
 
-| Scenario | What goes wrong | What OCW catches |
+| Scenario | What goes wrong | What TokenJam catches |
 |---|---|---|
 | [`retry-loop`](incidents/retry-loop/README.md) | Agent retries a failing tool in a loop, burning time and tokens | `retry_loop` + `failure_rate` alerts fire automatically |
 | [`surprise-cost`](incidents/surprise-cost/README.md) | Model silently escalates from Haiku to Opus mid-chain | Per-model cost breakdown shows the $3+ you didn't expect |
 | [`hallucination-drift`](incidents/hallucination-drift/README.md) | Agent behavior shifts — different tokens, different tools | `drift_detected` alert fires with Z-scores at session end |
 
-Each scenario runs against an in-memory backend and produces a side-by-side comparison: what `print()` shows vs. what OCW reveals.
+Each scenario runs against an in-memory backend and produces a side-by-side comparison: what `print()` shows vs. what TokenJam reveals.
 
 ---
 

{tokenjam-0.2.0 → tokenjam-0.2.1}/SECURITY.md

@@ -26,4 +26,4 @@ Security issues in these areas are especially important to report.
 ## Not in Scope
 
 - Vulnerabilities in upstream dependencies (report to the upstream project)
-- Issues that require physical access to the machine running `ocw`
+- Issues that require physical access to the machine running `tj`

{tokenjam-0.2.0 → tokenjam-0.2.1}/docs/alerts.md

@@ -1,6 +1,6 @@
 # Alerts
 
-`ocw` fires alerts the moment something happens — sensitive tool calls, budget breaches, behavioral drift, sandbox violations. Alerts are evaluated after every span ingest and dispatched to your configured channels in real time.
+`tj` fires alerts the moment something happens — sensitive tool calls, budget breaches, behavioral drift, sandbox violations. Alerts are evaluated after every span ingest and dispatched to your configured channels in real time.
 
 ## Alert types
 
@@ -22,7 +22,7 @@
 
 ## Channels
 
-Configure where alerts go in `.ocw/config.toml`. Multiple channels work simultaneously — you can get push notifications on your phone and a Discord message at the same time.
+Configure where alerts go in `.tj/config.toml`. Multiple channels work simultaneously — you can get push notifications on your phone and a Discord message at the same time.
 
 ```toml
 # Push notification (free, no account required)
@@ -49,7 +49,7 @@ url = "https://your-endpoint.com/alerts"
 # Local file log
 [[alerts.channels]]
 type = "file"
-path = "~/.ocw/alerts.log"
+path = "~/.tj/alerts.log"
 
 # Stdout (always enabled by default)
 [[alerts.channels]]
@@ -77,7 +77,7 @@ Define which tool calls should trigger immediate alerts:
 
 ## Cooldown
 
-To prevent alert storms, `ocw` tracks a cooldown per agent + alert type. Repeat alerts within the cooldown window are suppressed — still persisted to the database, but not dispatched to channels.
+To prevent alert storms, `tj` tracks a cooldown per agent + alert type. Repeat alerts within the cooldown window are suppressed — still persisted to the database, but not dispatched to channels.
 
 ```toml
 [alerts]

{tokenjam-0.2.0 → tokenjam-0.2.1}/docs/architecture.md

@@ -1,6 +1,6 @@
 # Architecture
 
-This document describes the architecture of TokenJam (ocw) — a local-first, OTel-native observability CLI for autonomous AI agents.
+This document describes the architecture of TokenJam (`tj`) — a local-first, OTel-native observability CLI for autonomous AI agents.
 
 ## Design principles
 
@@ -10,7 +10,7 @@ This document describes the architecture of TokenJam (ocw) — a local-first, OT
 
 3. **Never crash the agent.** The SDK, ingest pipeline, and all post-ingest hooks are designed to fail silently. Hook failures are logged but never propagated. A span rejection never takes down the instrumented agent.
 
-4. **Core is pure domain logic.** `ocw/core/` has no CLI or HTTP imports. CLI and API layers import from core, never the reverse. This makes the domain logic independently testable and reusable.
+4. **Core is pure domain logic.** `tokenjam/core/` has no CLI or HTTP imports. CLI and API layers import from core, never the reverse. This makes the domain logic independently testable and reusable.
 
 ---
 
@@ -42,7 +42,7 @@ flowchart TD
     Alerts --> DB
     Schema --> DB
 
-    DB --> CLI["ocw CLI"]
+    DB --> CLI["tj CLI"]
     DB --> API["REST API\n:7391/docs"]
     DB --> WebUI["Web UI\n:7391/"]
     DB --> MCP["MCP Server\nstdio"]
@@ -91,7 +91,7 @@ Sessions are identified by `conversation_id`. When a span arrives with a `conver
 ## Package structure and dependency rules
 
 ```
-ocw/
+tokenjam/
 ├── core/       Pure domain logic — NO imports from cli/ or api/
 ├── cli/        Click commands — imports from core/
 ├── api/        FastAPI routes — imports from core/
@@ -141,7 +141,7 @@ DuckDB allows only one write connection. When `tj serve` is running, CLI command
 
 ### Bootstrap
 
-`ensure_initialised()` in `ocw/sdk/bootstrap.py` is the lazy, thread-safe, idempotent entry point. It bootstraps: config loading → DB connection → IngestPipeline creation → TracerProvider setup. Called automatically by `@watch()` and all `patch_*()` functions. Registers an `atexit` handler to flush pending spans.
+`ensure_initialised()` in `tokenjam/sdk/bootstrap.py` is the lazy, thread-safe, idempotent entry point. It bootstraps: config loading → DB connection → IngestPipeline creation → TracerProvider setup. Called automatically by `@watch()` and all `patch_*()` functions. Registers an `atexit` handler to flush pending spans.
 
 ### @watch() creates sessions, not LLM spans
 
@@ -256,7 +256,7 @@ A stub endpoint at `/v1/metrics` returns 200 OK to prevent noisy warnings from O
 
 ## Web UI
 
-A single-file SPA (`ocw/ui/index.html`) served by `tj serve` at `http://127.0.0.1:7391/`. No build step, no node_modules — vanilla JS with Alpine.js from CDN.
+A single-file SPA (`tokenjam/ui/index.html`) served by `tj serve` at `http://127.0.0.1:7391/`. No build step, no node_modules — vanilla JS with Alpine.js from CDN.
 
 Six views: Status, Traces (with span waterfall visualization), Cost, Alerts, Budget, Drift. Hash-based routing (`/#/status`, `/#/traces`, etc.). The UI consumes the same REST API endpoints as external clients. All views auto-poll (Status at 5s, all others at 10s).
 
@@ -287,7 +287,7 @@ A fifth layer (`tests/e2e/`) makes real LLM API calls and is auto-skipped withou
 
 - **OTel TracerProvider:** Global and set-once per process. In SDK tests, set the provider once at module level (not per-test) and clear spans between tests using a custom `_CollectingExporter`.
 
-- **CLI tests:** Use `click.testing.CliRunner` with `unittest.mock.patch` on `ocw.cli.main.load_config` and `ocw.cli.main.open_db` to inject test fixtures.
+- **CLI tests:** Use `click.testing.CliRunner` with `unittest.mock.patch` on `tokenjam.cli.main.load_config` and `tokenjam.cli.main.open_db` to inject test fixtures.
 
 - **API tests:** Use `httpx.AsyncClient` with `httpx.ASGITransport(app=app)` against `InMemoryBackend`.
 
@@ -297,7 +297,7 @@ A fifth layer (`tests/e2e/`) makes real LLM API calls and is auto-skipped withou
 
 Config is TOML, discovered in order: `tj.toml` → `.tj/config.toml` → `~/.config/tj/config.toml`. Override with `--config` flag or `TJ_CONFIG` env var.
 
-The `TjConfig` dataclass tree in `ocw/core/config.py` defines the full hierarchy: agents (with budgets, sensitive actions, drift config, output schema), storage, export (OTLP, Prometheus), alerts (cooldown, channels), security, API, and capture settings.
+The `TjConfig` dataclass tree in `tokenjam/core/config.py` defines the full hierarchy: agents (with budgets, sensitive actions, drift config, output schema), storage, export (OTLP, Prometheus), alerts (cooldown, channels), security, API, and capture settings.
 
 `tomllib.load()` requires binary mode (`"rb"`) — text mode raises `TypeError` at runtime. The codebase uses conditional imports: `tomllib` (Python 3.11+) or `tomli` (3.10). Writing config uses `tomli_w`.
 
@@ -305,7 +305,7 @@ The `TjConfig` dataclass tree in `ocw/core/config.py` defines the full hierarchy
 
 ## MCP server (Claude Code integration)
 
-`tj mcp` is a stdio-based MCP (Model Context Protocol) server that gives Claude Code direct access to OCW observability data. It exposes 13 tools that Claude Code can call during a session.
+`tj mcp` is a stdio-based MCP (Model Context Protocol) server that gives Claude Code direct access to TokenJam observability data. It exposes 13 tools that Claude Code can call during a session.
 
 ### Dual-mode operation
 
@@ -348,7 +348,7 @@ If `tj serve` is not running, `cmd_mcp.py` can auto-start it as a detached subpr
 The `--claude-code` flag configures the full telemetry pipeline in one command:
 
 1. **Derives agent ID** from the git remote origin name (fallback: folder name), prefixed with `claude-code-`
-2. **Writes OCW config** to `~/.config/tj/config.toml` (global, shared across all projects) with agent entry and optional daily budget
+2. **Writes TokenJam config** to `~/.config/tj/config.toml` (global, shared across all projects) with agent entry and optional daily budget
 3. **Updates global Claude settings** (`~/.claude/settings.json`) with OTLP exporter env vars: `CLAUDE_CODE_ENABLE_TELEMETRY=1`, `OTEL_LOGS_EXPORTER=otlp`, endpoint, protocol. On re-runs, always resyncs the `OTEL_EXPORTER_OTLP_HEADERS` auth header to fix 401s without manual setup.
 4. **Writes project settings** (`./.claude/settings.json`) with `OTEL_RESOURCE_ATTRIBUTES=service.name={agent_id}` so spans are tagged to the right agent
 5. **Updates shell env** (`~/.zshrc`) with Docker-compatible endpoint (`host.docker.internal:{port}`) for harness sessions that can't reach `127.0.0.1`
@@ -373,7 +373,7 @@ The converted spans flow through the standard `IngestPipeline.process()` path
 
 ### Semantic conventions (`ClaudeCodeEvents`)
 
-`ocw/otel/semconv.py` defines `ClaudeCodeEvents` constants for Claude Code's log event attributes: session context (`session.id`, `prompt.id`, `event.sequence`), API request metrics (`cost_usd`, `duration_ms`, `speed`, token counts), tool result fields (`success`, `error`, `tool_parameters`), and error context (`status_code`, `attempt`).
+`tokenjam/otel/semconv.py` defines `ClaudeCodeEvents` constants for Claude Code's log event attributes: session context (`session.id`, `prompt.id`, `event.sequence`), API request metrics (`cost_usd`, `duration_ms`, `speed`, token counts), tool result fields (`success`, `error`, `tool_parameters`), and error context (`status_code`, `attempt`).
 
 ---
 
@@ -412,8 +412,8 @@ Z-scores are color-coded: green (|z| < 1.0), yellow (approaching threshold), red
 
 ## Packaging
 
-- **PyPI package name:** `tokenjam` (not `ocw`)
-- **CLI command:** `ocw`
-- **Python package directory:** `ocw/`
-- **Build system:** hatchling, with `[tool.hatch.build.targets.wheel] packages = ["ocw"]` because the package name differs from the directory name
+- **PyPI package name:** `tokenjam`
+- **CLI command:** `tj`
+- **Python package directory:** `tokenjam/`
+- **Build system:** hatchling, with `[tool.hatch.build.targets.wheel] packages = ["tokenjam"]`
 - **npm package:** `@tokenjam/sdk` under `sdk-ts/`

{tokenjam-0.2.0 → tokenjam-0.2.1}/docs/claude-code-integration.md

@@ -10,7 +10,7 @@ tj status --agent claude-code-<project>
 ```
 
 `tj onboard --claude-code` does everything in one shot:
-- Creates a shared config at `~/.config/ocw/config.toml` (one config for all your projects)
+- Creates a shared config at `~/.config/tj/config.toml` (one config for all your projects)
 - Writes OTLP exporter vars to `~/.claude/settings.json` so Claude Code sends telemetry automatically
 - Tags this project's sessions by writing `OTEL_RESOURCE_ATTRIBUTES=service.name=claude-code-<project>` to `.claude/settings.json`
 - Registers the MCP server globally (`claude mcp add --scope user tj -- tj mcp`)
@@ -50,32 +50,32 @@ The MCP server is included in the `[mcp]` extra and registered automatically by
 | `get_tool_stats` | Tool call counts and average duration |
 | `get_drift_report` | Behavioral drift baseline vs latest session |
 | `acknowledge_alert` | Mark an alert as acknowledged |
-| `setup_project` | Configure a project to send telemetry to OCW |
+| `setup_project` | Configure a project to send telemetry to TokenJam |
 | `open_dashboard` | Open the web UI — starts `tj serve` on demand if needed |
 
 The MCP server opens the DuckDB file read-only — no lock conflicts with `tj serve` if both are running. The single write operation (`acknowledge_alert`) opens a short-lived read-write connection only for its UPDATE.
 
 **Per-project telemetry tagging** — after installing the MCP server globally, ask Claude Code to set up each project:
 
-> "Set up OCW for this project"
+> "Set up TokenJam for this project"
 
 Claude calls `setup_project`, which writes `.claude/settings.json` with `OTEL_RESOURCE_ATTRIBUTES=service.name=<project>` so spans from that project are tagged with the right agent ID.
 
 ## Uninstalling
 
 ```bash
-# Remove all OCW data, config, daemon, MCP registration, and env vars from every onboarded project:
+# Remove all TokenJam data, config, daemon, MCP registration, and env vars from every onboarded project:
 tj uninstall --yes
 
-# Then remove the package itself (ocw uninstall intentionally skips this):
+# Then remove the package itself (tj uninstall intentionally skips this):
 pip uninstall tokenjam -y
 ```
 
 `tj uninstall` cleans up everything set by `tj onboard --claude-code`:
 - Stops and removes the background daemon (launchd/systemd)
 - Deregisters the MCP server from Claude Code
-- Deletes `~/.ocw/` (telemetry database)
-- Deletes `~/.config/ocw/` (global config and projects index)
+- Deletes `~/.tj/` (telemetry database)
+- Deletes `~/.config/tj/` (global config and projects index)
 - Removes OTLP env vars from `~/.claude/settings.json`
 - Removes `OTEL_RESOURCE_ATTRIBUTES` from `.claude/settings.json` in every onboarded project
 - Removes the harness env block from `~/.zshrc`

{tokenjam-0.2.0 → tokenjam-0.2.1}/docs/cli-reference.md

@@ -156,7 +156,7 @@ tj stop
 
 ### `tj uninstall`
 
-Remove all OCW data, config, daemon, MCP registration, and env vars.
+Remove all TokenJam data, config, daemon, MCP registration, and env vars.
 
 ```bash
 tj uninstall          # interactive confirmation

{tokenjam-0.2.0 → tokenjam-0.2.1}/docs/configuration.md

@@ -1,6 +1,6 @@
 # Configuration
 
-`ocw` uses TOML configuration. Run `tj onboard` to generate a starter config, or create one manually.
+`tj` uses TOML configuration. Run `tj onboard` to generate a starter config, or create one manually.
 
 ## Config file discovery
 
@@ -45,11 +45,11 @@ completions  = false
 tool_outputs = false
 
 [storage]
-path           = "~/.ocw/telemetry.duckdb"
+path           = "~/.tj/telemetry.duckdb"
 retention_days = 90
 
 [security]
-ingest_secret = "ocw_..."   # generated by tj onboard
+ingest_secret = "..."   # generated by tj onboard
 
 [api]
 host = "127.0.0.1"
@@ -90,7 +90,7 @@ Set limits via CLI (`tj budget --daily 10`), the REST API (`POST /api/v1/budget`
 
 ## Capture settings
 
-By default, `ocw` does not capture prompt content, completion content, or tool outputs. Enable these selectively if you need them for debugging or evaluation:
+By default, `tj` does not capture prompt content, completion content, or tool outputs. Enable these selectively if you need them for debugging or evaluation:
 
 ```toml
 [capture]

{tokenjam-0.2.0 → tokenjam-0.2.1}/docs/export.md

@@ -1,6 +1,6 @@
 # Export and Integrate
 
-`ocw` stores all telemetry in a local DuckDB database. Export it in multiple formats for analysis, evaluation, or forwarding to external backends.
+`tj` stores all telemetry in a local DuckDB database. Export it in multiple formats for analysis, evaluation, or forwarding to external backends.
 
 ## Export formats