UsageSniffer
Scan your local Claude Code, Codex, Opencode, Gemini CLI, Copilot CLI, Cursor, Aider, and Continue sessions and see visually what eats your tokens — skills, thinking, tools, context cache, plus dollar costs, anomalies, and cross-agent comparisons.
Zero-dependency Python. Your session logs never leave the machine.
Install
Requires Python 3.10+.
# run straight from the repo (no install needed)
python3 -m usagesniffer scan --top 10
# or install it so the `usagesniffer` command is on your PATH
pip install git+https://github.com/Sleepy-Studio/UsageSniffer.git
# or: pipx install git+https://github.com/Sleepy-Studio/UsageSniffer.git
usagesniffer scan --top 10
Quick start
cd UsageSniffer
python3 -m usagesniffer scan --top 10 # everything
python3 -m usagesniffer scan --agents claude --top 5
python3 -m usagesniffer session 05062c81 # drill into one session (prefix match)
python3 -m usagesniffer cost --top 10 # dollar rollup + cache savings
python3 -m usagesniffer cost --budget 50 # exit 2 + alert when over $50
python3 -m usagesniffer report -o report.html
python3 -m usagesniffer watch --interval 30 # live burn (Ctrl-C to stop)
python3 -m usagesniffer anomalies # outlier sessions vs agent baseline
python3 -m usagesniffer skills-roi # per-skill cost table
python3 -m usagesniffer compare # cross-agent efficiency
# filters work on most commands:
python3 -m usagesniffer scan --since 2026-09-01 --project Sunrise --model sonnet
What it reads
| Agent | Source | Ground-truth token signal |
|---|---|---|
| Claude | ~/.claude/projects/*/*.jsonl |
message.usage: input / output / cache_creation / cache_read per assistant message |
| Codex | ~/.codex/sessions/**/*.jsonl |
event_msg → token_count: last_token_usage per step (input / cached / output / reasoning) |
| Opencode | ~/.local/share/opencode/opencode.db |
message.data.tokens + part rows (text / reasoning / tool) |
| Gemini | ~/.gemini/tmp/*/chats/*.json |
per-message tokens {input, output, cached, thoughts, tool} |
| Copilot | ~/.copilot/session-store.db |
assistant_usage_events per turn (input / output / cache / reasoning). Subscription billing → $ unknown |
| Cursor | ~/.config/Cursor/User/**/state.vscdb |
⚠️ per-bubble tokens read as zero on current builds; char-estimated, cost from usageData.costInCents when present |
| Aider | .aider.chat.history.md / --analytics-log JSONL |
real tokens only via analytics log (prompt_tokens, completion_tokens, cost); transcripts are char-estimated |
| Continue | ~/.continue/sessions/*.json |
session usage {promptTokens, completionTokens, cachedTokens} when the model reports it |
Attribution buckets
Real totals always come straight from the logs. The buckets are token-weighted estimates (labelled as such in output):
thinking— Claude thinking blocks / Codexreasoning_output_tokens/ Opencode reasoning partsskills—SKILL.mdloads, skill-instruction blocks, Skill-tool callstools— tool_use + tool_result / exec payload sharecontext_cache/context_write— cache read / creation tokensassistant_text,system,user,other
The second chart ("action share EXCLUDING context cache") is the useful one: cache reads dominate raw totals but are cheap; the action chart shows where your real budget goes.
Layout
usagesniffer/
models.py # SessionRecord / ScanResult
analyze.py # aggregation
render.py # terminal ASCII charts (stdlib only)
cli.py # scan | top | session
parsers/
claude.py # ~/.claude/projects JSONL
codex.py # ~/.codex/sessions rollout JSONL
opencode.py # opencode.db SQLite
Roadmap
- per-model pricing + $ cost rollup (
cost,--budget) -
--watchlive mode - static HTML report export (
report) - prompt-caching savings estimate (cache_read vs full-price replay)
- 8-agent coverage (Claude, Codex, Opencode, Gemini, Copilot, Cursor, Aider, Continue)
- anomalies, skill ROI, cross-agent compare
- PyPI publish (
pipx install usagesniffer) — seepackaging.md - Homebrew / AUR packages — templates in
packaging.md
1 maintainer
Metadata
1 maintainer
Release files for usagesniffer 1.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| usagesniffer-1.1.1.tar.gz | 25.3 kB | With provenance | Details for usagesniffer-1.1.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| usagesniffer-1.1.1-py3-none-any.whl | Python 3 | none | any | Details for usagesniffer-1.1.1-py3-none-any.whl |
Total release size: 57.3 kB
Release files / usagesniffer-1.1.1.tar.gz
| Download URL | usagesniffer-1.1.1.tar.gz |
|---|---|
| Size | 25.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8ed558f3a6e5321292638af1ebdae948fa4b7a5e8c936f10ad8810c8cfe94189
|
|
BLAKE2b-256 checksum How to use checksums |
e1e1701b923297929aa42dd0d0378fce5bbc90e4104c3c669d832bb7dfa52b21
|
| 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 17, 2026.
Transparency logRelease files / usagesniffer-1.1.1-py3-none-any.whl
| Download URL | usagesniffer-1.1.1-py3-none-any.whl |
|---|---|
| Size | 32.0 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
f3fa3be75a4ab314264a2e8b43b199b9e9bcfb626a4b599f2016add0d61bb56a
|
|
BLAKE2b-256 checksum How to use checksums |
eed0ea87f5ae030034235948ad1379021541d251d67934ff64caddec3283c098
|
| 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 17, 2026.
Transparency logSecurity for usagesniffer 1.1.1
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.
Full provenance
All 2 files in this release have been uploaded with provenance metadata.
Attestation for usagesniffer-1.1.1.tar.gz
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 17, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
