`claude -p --output-format json` shares stdout with the CLI's own
subsystems, so the envelope is not always the first thing on it. An
attached MCP server that advertises no tools makes every invocation emit
Client.listTools() called but server does not advertise tools
capability - returning empty list
ahead of the envelope, and `json.loads(stdout)` fails on the whole buffer.
The failure is raised after the model has already answered, so the chunk
is discarded with its tokens spent. On a real corpus this burned 149,582
input and 52,162 output tokens on chunk 1 of 3 and returned nothing, and
the error names the JSON rather than the preamble that caused it, so the
log points at the wrong thing. Anyone with an MCP server configured hits
it on every chunk.
`_envelope_after_preamble` scans for the first `[`/`{` that begins a
valid JSON document. `raw_decode` ignores trailing bytes, so a diagnostic
on either side is tolerated. The fast path is unchanged: clean stdout
still parses on the first `json.loads`, and stdout with no JSON at all
still returns None so the caller raises -- test_raises_on_garbage_envelope
holds.
(cherry picked from commit ede3a9e0a8)
#2543 (adopts PR #2546, thanks @michaelxer): a failed extraction is no
longer stamped in the incremental manifest as up-to-date, so graphify
update retries it instead of skipping it forever; a manifest already
poisoned by the old behavior is healed on the next run; genuinely
unchanged files are not re-processed. Extended to the watch save_manifest
paths too.
#2554 (adopts PR #2555, thanks @annieyii): the claude-cli backend now
inspects the stdout envelope for an is_error result (e.g. a rate limit
returned with exit code 0) and raises it on both the zero and non-zero
exit paths, instead of parsing it as an empty success and bisecting
against a live rate limit.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The claude-cli backend delivers the extraction schema in the user turn and
trusts the model to emit raw JSON. Newer Claude Code releases treat that
prompt as an agentic task and report the result in prose instead ("Knowledge
graph extracted — 21 nodes, 20 edges…"), so the graph parses empty, reads as
truncation, and adaptive-retry bisects without ever converging.
Pass --json-schema (structured output) when the CLI advertises it — probed
once via `claude --help` and cached — so the object shape is constrained
regardless of prompt framing. Older CLIs that predate the flag keep the
user-turn prompt as a fallback. The `result` envelope still carries the JSON
string, so the parse path is unchanged.
The claude-cli backend passed the extraction schema via --system-prompt with
only the raw file dump in the user turn, assuming a replacement system prompt
is the model's sole authority. Claude Code >= ~2.1 (verified on 2.1.197) does
not honour that: it still layers in the local coding-agent context
(CLAUDE.md/AGENTS.md in cwd, skills, MCP) and, given a user turn that is just a
file with no request, replies conversationally ("I see the file, but there's no
actual request attached"). That prose parses to zero nodes/edges, so
_response_is_hollow flags it as truncation and the adaptive-retry path bisects
the chunk indefinitely (94 -> 47 -> 23 -> ...), never converging and never
writing graph.json.
Move the full extraction schema plus an explicit imperative into the user turn
and drop --system-prompt, so the CLI emits the JSON object directly. The
<untrusted_source> prompt-injection guardrails are carried verbatim; model
override, --add-dir image handling, timeout, and token accounting are untouched.
Two Windows fixes for the claude-cli backend:
1. _call_llm (the label/community-naming path) spawned a bare ["claude", ...],
which CreateProcess cannot resolve to the npm claude.cmd shim (PATHEXT does
not apply) — every labeling batch failed with WinError 2. Mirror the
extraction path's resolution: prefer shutil.which("claude.cmd") and pass
the resolved path. Regression test included.
2. Both claude -p spawn sites now pass CREATE_NO_WINDOW on Windows. Without
it, each per-batch claude.cmd spawn allocates a console — with Windows
Terminal as the default terminal, a labeling run pops one visible window
per batch on the user's desktop for the duration of the model call.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
prav