headroom/headroom
Tejas Chopra 4f2e70a75c
fix(proxy/anthropic): authenticate and attribute buffered Copilot turns (#3277)
## Description

Follow-up to #3258. That PR points the Anthropic target at the Copilot
host so Claude models stop 401'ing. This PR fixes two things on the
Anthropic path that were only ever correct on the **streaming** arm, and
which #3258 makes reachable for real Copilot traffic.

Copilot serves Claude models from its Anthropic surface (`/v1/messages`)
on the same host as its OpenAI surface, so the resolved Anthropic target
can be a Copilot host with no per-request `upstream_base_url` involved.
That is the case both arms below get wrong.

**1. The buffered arm sent no Copilot credential.**
`apply_copilot_api_auth` is keyed on the upstream URL and was applied
only by `_stream_response` (`handlers/streaming.py:1205`). The
buffered/non-stream arm sends through `_retry_request`
(`proxy/server.py:2132`), which forwards headers untouched — so the
request carried whatever the client happened to send and none of
Headroom's own credential handling: no minted or refreshed token (the
one `wrap vscode` explicitly hands the proxy), no
`Copilot-Integration-Id` default. A client token that went stale
mid-session 401'd here while the streaming path recovered. That arm is
not an edge case — it is the CCR `stream:true → buffered stream:false`
flip, and Claude Code's non-stream retry.

**2. Copilot turns were attributed to "anthropic".**
`build_copilot_upstream_url` is the only place
`mark_request_routed_to_copilot` fires (`copilot_auth.py:1288`), and
`emit_request_outcome` relabels the provider off that flag
(`proxy/outcome.py:419`). The buffered arm built its URL by f-string,
skipping the chokepoint, so those turns showed as `anthropic` on the
dashboard. The URL produced is byte-identical either way — this is
attribution only, not routing. `proxy/cost.py` has no Copilot-specific
branch, so pricing is unaffected.

Both changes are inert off the Copilot path: `apply_copilot_api_auth`
returns the headers unchanged for a non-Copilot URL, and
`build_copilot_upstream_url` only joins base + path there.

Independent of #3258 and based on `main` — the gaps are reachable today
by setting `ANTHROPIC_TARGET_API_URL` to a Copilot host.

## Type of Change

- [x] Bug fix (non-breaking change that fixes an issue)

## Changes Made

- `handlers/anthropic.py`: build the default-target URL through
`build_copilot_upstream_url` instead of an f-string, so the
routed-to-Copilot flag is set for attribution.
- `handlers/anthropic.py`: apply `apply_copilot_api_auth` on the
buffered arm before the upstream send. Mutated in place, matching the
accept-header handling directly above — the closures below capture
`headers`, and the CCR continuation rebuilds its own header set from it,
so the continuation inherits the auth too.
- New test pinning both at the `_retry_request` seam: URL built, headers
as they go on the wire, and the flag as it stands at send time.

## Testing

- [x] Unit tests pass (`pytest`)
- [x] Linting passes (`ruff check`, CI-pinned 0.16.3)
- [x] Type checking passes (`mypy headroom`)
- [x] New tests added for new functionality

### Test Output

Both new assertions fail on `main` with exactly the symptoms described,
and pass with the fix:

```text
$ git stash && pytest tests/test_proxy/test_anthropic_copilot_upstream_auth.py
tests/.../test_buffered_turn_to_copilot_is_authenticated
E   KeyError: 'authorization'
tests/.../test_buffered_turn_to_copilot_is_flagged_for_attribution
E   assert False is True
==================== 2 failed, 2 passed, 1 warning in 3.38s ====================

$ git stash pop && pytest tests/test_proxy/test_anthropic_copilot_upstream_auth.py
========================= 4 passed, 1 warning in 2.88s =========================
```

The two that pass on `main` are the invariants this must not break (path
`/v1` preserved per #2409, non-Copilot target untouched).

Regression run over the affected surface:

```text
$ pytest tests/ -k "copilot or anthropic or outcome or provider_registry or proxy_routes or upstream"
= 3 failed, 1111 passed, 33 skipped, 11112 deselected in 152.98s =
```

The 3 failures are
`tests/test_proxy/test_openai_transport_path_prefix.py` and are
**pre-existing on `main`** (verified by running that file on a clean
checkout — same 3 fail). Untouched by this PR, which is Anthropic-path
only.

```text
$ uvx ruff@0.16.3 check headroom/proxy/handlers/anthropic.py tests/test_proxy/test_anthropic_copilot_upstream_auth.py
All checks passed!
$ mypy headroom/proxy/handlers/anthropic.py
Success: no issues found in 1 source file
```

## Real Behavior Proof

- **Environment:** macOS arm64, Python 3.12.13, `main` @ 0.36.5.
- **Exact command / steps:** drive `POST /v1/messages` through the real
app (`create_app` + `TestClient`, non-stream body) with the Anthropic
target set to `https://api.githubcopilot.com`, intercepting
`_retry_request` to capture what was about to go on the wire. Copilot
token minting stubbed to a fixed value.
- **Observed result:** before — no `Authorization` header at all on the
buffered arm, and `request_routed_to_copilot()` is `False` at send time.
After — `Authorization: Bearer <minted>` plus `Copilot-Integration-Id`
and `Editor-Version`, flag `True`, URL unchanged at
`https://api.githubcopilot.com/v1/messages`. With a non-Copilot target,
no credential is invented and the flag stays `False`.
- **Not tested:** against live `api.githubcopilot.com` — no Copilot
subscription in this environment. Token minting is stubbed, so the
refresh path itself is exercised only to the provider boundary.
Anthropic **batch** endpoints (`/v1/messages/batches`,
`handlers/anthropic.py:5066+`) still build against
`self.ANTHROPIC_API_URL` and will point at Copilot, which does not serve
them — pre-existing and out of scope here — filed as #3278.

## Runtime Rollout Safety

- **Rollout-managed feature(s):** none — no flag or channel involved.
- **Minimum rollout channel:** n/a.
- **Stable/default behavior changed:** no, for every non-Copilot
upstream: the URL is byte-identical and `apply_copilot_api_auth`
early-returns for non-Copilot URLs. Behavior changes only when the
Anthropic target is a Copilot host, which is the broken case.
- **Kill switch / disable path:** set `ANTHROPIC_TARGET_API_URL` to a
non-Copilot host; both paths go inert.
- **Unsafe override required:** none.
- **Qualification impact:** none.
- **Rollback path:** revert this commit — it is self-contained to one
file plus a new test.

## Review Readiness

- [x] I have performed a self-review
- [x] This PR is ready for human review

---------

Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
2026-08-26 23:44:03 +05:30
..
audit fix: remove rtk and lean-ctx CLI context tools (#2677) 2026-07-30 22:59:41 -07:00
backends fix(proxy): pass through cross-region prefixed Bedrock model IDs directly (#2330) 2026-08-11 23:54:03 -05:00
cache feat(proxy): unify proxy and sidecar compression on one session engine (#3271) 2026-08-26 15:57:31 +05:30
capture fix(cli): harden all CLI surfaces + fix docs accuracy (#1491) 2026-06-27 14:48:43 -07:00
ccr fix(ccr): make StreamingCCRHandler work on OpenAI streams (#3069) 2026-08-17 20:55:02 -07:00
cli fix(copilot): preserve native enterprise model routing (#2998) 2026-08-25 21:46:31 -05:00
compression fix(code): parse-probe tree-sitter availability in code_handler (#1231) (#1300) 2026-07-09 14:06:29 -05:00
dashboard fix(dashboard): pin MIME types for the vendored static assets (#3193) 2026-08-21 22:28:26 -07:00
evals feat(evals): weekly HotpotQA answer-recall report on the prose path (#1188) 2026-07-15 21:40:55 +00:00
graph fix(security): address u9up assessment findings (WEB-01–07) (#2207) 2026-08-20 09:02:44 -05:00
image fix(image): decouple routing types from trained_router so importing the compressor doesn't import torch (#2513) (#2537) 2026-08-12 00:22:56 -05:00
install fix(wrap): set xAI upstream for grok-build proxy (#2772) 2026-08-16 15:04:59 -07:00
integrations fix(litellm): close shared cloud client 2026-08-11 10:13:10 -07:00
learn fix(learn): surface Codex analysis failures (#3016) 2026-08-25 21:40:12 -05:00
mcp_registry fix(mcp): add explicit Serena reconciliation (#3222) 2026-08-23 11:52:50 -07:00
memory fix(memory/graph): skip a corrupt row instead of aborting a whole graph scan (#3239) 2026-08-25 11:40:54 +05:30
models fix(models): version-boundary longest-prefix match in ModelRegistry.get (#1658) 2026-07-10 23:07:31 -05:00
observability fix(ci): prevent native detector from hanging test shards (#2996) 2026-08-13 20:47:55 -05:00
perf fix(metrics): attribute tool-schema savings per model, not just compression (#3155) 2026-08-20 11:54:02 -07:00
pricing fix(anthropic): honor the [1m] 1M-context tier, and price it correctly (#3073) 2026-08-16 22:46:12 -07:00
providers fix(providers): route Claude requests to Copilot when the OpenAI target is a Copilot host (#3258) 2026-08-26 22:43:11 +05:30
proxy fix(proxy/anthropic): authenticate and attribute buffered Copilot turns (#3277) 2026-08-26 23:44:03 +05:30
relevance feat(security): pilot hardening — stateless guarantee, model pinning, CI security gate (#1515) 2026-06-27 17:44:10 -07:00
reporting feat: attribute reread waste to over-compression via marker check (#901) 2026-06-13 10:43:35 -05:00
storage Fix 19 test failures: missing security attr, nosec inside f-strings, stale test mock 2026-04-07 18:30:29 -07:00
subscription fix(subscription): dedup transcript usage by message id (#2340 token inflation) (#2408) 2026-08-12 00:05:04 -05:00
telemetry fix(proxy/metrics): cap client-supplied model label cardinality (#2480) 2026-08-12 00:15:49 -05:00
testing deps: bump ruff from 0.15.22 to 0.16.2 in the pip-minor-patch group across 1 directory (#2962) 2026-08-14 16:38:08 -05:00
tokenizers perf: cut hot-path latency 27% (token-count memo, startup preloads, JSON scan memo) (#2838) 2026-08-06 17:47:40 -07:00
transforms feat(proxy): self-limiting session state for the compression-cache registry (#3261) 2026-08-26 15:04:43 +05:30
__init__.py Pin ORT dylib on Windows; init Python logging (#1010) 2026-06-23 07:46:24 -05:00
_ort.py fix(onnx): enforce Rust API-24 runtime compatibility (#2979) 2026-08-13 15:05:41 -05:00
_subprocess.py ci: repair mypy no-any-return in _win32_pid_alive (#1556 follow-up) (#2336) 2026-07-16 20:55:17 -07:00
_version.py fix(version): mark source-checkout builds as -dev (#2072) 2026-07-13 09:38:17 -04:00
agent_savings.py fix(reporting): show net vs gross savings, real skip thresholds, and the effective profile (#3123) 2026-08-19 00:16:37 -07:00
binaries.py fix(security): address u9up assessment findings (WEB-01–07) (#2207) 2026-08-20 09:02:44 -05:00
cli.py Add Click-based CLI with memory management commands 2026-01-29 21:30:21 -08:00
client.py refactor: DRY cache logic, add thread safety, fix Bash exclusion (#704) 2026-06-16 14:50:04 -05:00
compress.py feat(compress): expose frozen_message_count in library-mode compress() (#2178) 2026-07-14 16:07:21 -04:00
config.py fix(proxy): protect file reads from lossy compression on the Responses API path (Copilot view + HEADROOM_PROTECT_READS) (#3238) 2026-08-24 19:03:03 +08:00
context_tool_cleanup.py fix(install): consolidate Windows fallback and cleanup safety (#2980) 2026-08-13 15:05:45 -05:00
copilot_auth.py fix(copilot): honor corporate TLS for token refresh (#3246) 2026-08-25 21:37:12 -05:00
copilot_linux_secret.py fix(windows): pin UTF-8 encoding on text-mode subprocess calls (#1311) 2026-06-23 12:52:49 -05:00
copilot_macos_keychain.py fix(windows): pin UTF-8 encoding on text-mode subprocess calls (#1311) 2026-06-23 12:52:49 -05:00
exceptions.py fix: B1 — retire ICM, RollingWindow, scoring, relevance + dependents 2026-05-02 12:23:17 -07:00
fsutil.py fix: remove rtk and lean-ctx CLI context tools (#2677) 2026-07-30 22:59:41 -07:00
hooks.py feat: introduce canonical pipeline lifecycle contract 2026-04-21 23:54:28 -05:00
offline.py feat(proxy): pilot hardening — inbound auth, security headers, audit log, air-gap switch (#1537) 2026-06-28 12:09:00 -07:00
onnx_runtime.py fix(onnx): stop ONNX thread pools from spinning idle cores (#2495) (#2540) 2026-08-08 01:33:57 -05:00
parser.py fix(agno): tolerate streaming tool-call SDK objects in parser (#1312) (#1336) 2026-06-24 09:52:15 -05:00
paths.py fix(wrap): serialize shared proxy startup (#2946) 2026-08-12 12:50:09 -07:00
pipeline.py feat: introduce canonical pipeline lifecycle contract 2026-04-21 23:54:28 -05:00
py.typed Prepare for OSS release v0.2.0 2026-01-07 11:36:44 -08:00
release_version.py ci(release): publish win_amd64 wheel so Windows installs need no Rust (#1328) (#1335) 2026-06-24 09:48:37 -05:00
rollout.py feat: add deterministic runtime rollout controls (#1490) 2026-08-12 23:16:54 -05:00
savings_ledger.py fix(savings): don't bill free models at the $3/M fallback in the ledger (#2147) 2026-07-14 12:19:30 -04:00
settings_store.py fix(ccr): resolve <<ccr:...>> markers inline when no retrieve-tool path exists (#2512) 2026-08-12 00:18:43 -05:00
shared_context.py fix(shared_context): don't evict an unrelated entry on an update at capacity (#2136) 2026-07-13 19:57:32 -04:00
tokenizer.py Initial commit: Headroom SDK - LLM context optimization toolkit 2026-01-06 23:16:58 -08:00
tools.json fix(security): address u9up assessment findings (WEB-01–07) (#2207) 2026-08-20 09:02:44 -05:00
update_check.py feat(proxy): pilot hardening — inbound auth, security headers, audit log, air-gap switch (#1537) 2026-06-28 12:09:00 -07:00
utils.py chore: add nosec B324 annotations to non-cryptographic MD5 usages and update temporary database path to use system temp directory 2026-04-07 13:07:26 +06:00