Adds dir_name field to SkillSummary to enable detection of skills where
the SKILL.md frontmatter name differs from the parent directory name.
Also adds SkillMetadataDrift struct for tracking mismatches (used by
skills JSON output in follow-up).
Generated with https://github.com/Yeachan-Heo/gajae-code
Co-authored-by: Gajae Code <dev@gajae-code.com>
claw --resume now enforces the same broad-cwd safety policy as claw prompt
and the interactive REPL. Running from /, $HOME, or other broad directories
blocks execution unless --allow-broad-cwd is passed.
Generated with https://github.com/Yeachan-Heo/gajae-code
Co-authored-by: Gajae Code <dev@gajae-code.com>
claw acp serve now exits 2 (not implemented) instead of 0, so automation
pipelines can detect the no-op via exit code gating.
Key changes:
- acp serve exits 2 instead of 0
- Removed discoverability_tracking, tracking, recommended_workflows from JSON
- Removed phase, exit_code, serve_alias_only fields from JSON
- Status changed from unsupported/discoverability_only to not_implemented
- Error kind for unsupported ACP invocations uses typed prefix
- Updated tests to match new exit code and JSON structure
Generated with https://github.com/Yeachan-Heo/gajae-code
Co-authored-by: Gajae Code <dev@gajae-code.com>
Agent discovery now loads .md files with YAML frontmatter alongside .toml
files, matching the Claude Code agent definition convention. Markdown
agent files must have ----delimited YAML frontmatter with at least name
or description fields.
Key changes:
- parse_agent_frontmatter extracts name, description, model, model_reasoning_effort
- load_agents_from_roots_with_invalids collects both valid and invalid agents
- InvalidAgentConfig tracks rejected .md files with reason
- AgentCollection groups valid agents with invalid entries
- agents JSON output includes valid_count, invalid_count, invalid_agents
- Status is degraded when invalid agents exist
Generated with https://github.com/Yeachan-Heo/gajae-code
Co-authored-by: Gajae Code <dev@gajae-code.com>
Hook config now supports the Claude Code structured hook format with
partial validation. Invalid hook entries are recorded in invalid_hooks
while valid siblings are retained, following the same pattern as MCP
partial validation (#440).
Key changes:
- RuntimeInvalidHookConfig now includes typed kind field (invalid_hooks_config
or unknown_hook_event) for machine-readable error classification
- Hook parsing collects all invalid entries instead of halting at first error
- Unknown hook event names recorded as invalid without rejecting valid hooks
- Legacy bare-string hooks still load with deprecation warnings
- Claude Code documented format loads without error (matcher + nested hooks)
- config/status/doctor JSON surfaces hook_validation metadata
- classify_error_kind maps hook errors to invalid_hooks_config
Generated with https://github.com/Yeachan-Heo/gajae-code
Co-authored-by: Gajae Code <dev@gajae-code.com>
- Fix latest_session_alias_resolves_most_recent_managed_session test:
the test created sessions with 0 messages, which are now filtered out
by the message_count > 0 check in latest_session_excluding(). Updated
the test to call push_user_text() before saving so sessions have
at least one message and are findable by /resume latest.
- Add distinct error message when all sessions are empty (0 messages).
Previously, the same "no managed sessions found" message was returned
whether there were zero sessions or all sessions had 0 messages. Now:
- No sessions at all → "no managed sessions found in {path}. Start
claw to create a session..."
- Sessions exist but all empty → "all sessions are empty (0 messages)
in {path}. This usually means a fresh claw session is running but
no messages have been sent yet. Wait for a response in your other
session, then try --resume latest again."
- Add test for the all-sessions-empty error path.
Addresses reviewer feedback on #3216.
- Add missing retry_after: None field to ApiError::Api construction
in main.rs test. This field was introduced by the Retry-After
header support but was not added to the test's error initializer,
causing a compile error under CI's strict mode.
- Remove duplicate #[must_use] attribute on retry_after() method
in error.rs (lines 134+138 both had it; kept the outer one
above the doc comment per convention).
- Cargo fmt --all run.
- Reviewer question "Are defaults preserved?" — answered yes:
ApiTimeoutConfig defaults to 30s connect / 300s request / 8 retries.
with_retry_policy() is opt-in. No behavior change without explicit
configuration.
Three improvements to the /resume command:
1. /resume latest now skips the current empty session
When a new session is created on startup (with 0 messages), /resume
latest previously returned that empty session. Now it skips sessions
with message_count == 0 and excludes the current session ID via the
new exclude_id parameter, so it finds the previous session with
actual conversation history.
2. Unified load_session_excluding() replaces load_session_loose()
The previous load_session_loose() only handled cross-workspace
resume for aliases. The new load_session_excluding() combines the
loose workspace validation logic with the exclude_id parameter,
simplifying the call chain and ensuring all resume paths skip the
current empty session when appropriate.
3. All existing session scanning paths (global root + project-local
.claw/sessions/) are already in place from prior commits, and now
the exclude_id filter is applied consistently across both local
and global session scans.
Changes:
- session_control.rs: Add resolve_reference_excluding() that delegates
from resolve_reference(), adding optional exclude_id filtering for
alias references.
- session_control.rs: Add latest_session_excluding() that delegates
from latest_session(), filtering out excluded session IDs and
sessions with 0 messages in both local and global scan paths.
- session_control.rs: Add load_session_excluding() that replaces
load_session_loose(), combining cross-workspace alias handling with
the exclude_id parameter.
- main.rs: Add load_session_reference_excluding() that delegates from
load_session_reference(), using the new store method.
- main.rs: Wire LiveCli::resume_session() to pass the current session
ID as the exclude_id so /resume latest skips the current empty
session.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Cherry-picked from PR #2816 onto current upstream/main, resolving
conflicts from PR #3015's merge (which added retry_after to ApiError
but some construction sites were missing it).
Commits preserved:
- ade85398: API timeout config, Retry-After header, configurable retry
- TimeoutConfig in HTTP client builder (connect 30s, request 5min)
- CLAW_API_CONNECT_TIMEOUT and CLAW_API_REQUEST_TIMEOUT env vars
- Retry-After header parsing on 429 responses
- ApiTimeoutConfig in runtime config (settings.json)
- 8a883430: retry 400 responses with transient gateway error bodies
- Detects known gateway phrases in 400 response bodies
- Marks them as retryable instead of hard-failing
- ed91a61e: add 'no parseable body' to CONTEXT_WINDOW_ERROR_MARKERS
- Some providers return 400 with 'no parseable body' for oversized
requests instead of a proper context_length_exceeded error
Commits skipped (already in upstream via PR #3015):
- 453ab642: optional id field (already merged)
- baa8d1ba: HTML detection in streaming (already merged)
- 33d2f789: JSON error detection in streaming (already merged)
8 files changed, 299 insertions, 80 deletions
Some OpenAI-compat backends (e.g. glm-5.1-fast) return 400 with
"no parseable body" when the request payload is too large to parse,
rather than a proper context_length_exceeded error. Without this marker,
is_context_window_error() returns false and the auto-compact retry
loop never triggers — the user just sees an opaque 400 error.
💘 Generated with Crush
Assisted-by: GLM 5.1 FP8 via Crush <crush@charm.land>
Some providers/proxies return HTTP 400 with bodies like "no parseable
body" or "connection reset" during transient network blips. These are
not real bad requests — they're gateway errors wearing a 400 mask.
Detect known gateway error phrases in 400 response bodies and mark
them as retryable so the existing exponential backoff handles them.
- Add TimeoutConfig to HTTP client builder with connect_timeout (30s)
and request_timeout (5min) defaults, configurable via
CLAW_API_CONNECT_TIMEOUT and CLAW_API_REQUEST_TIMEOUT env vars
- Add with_timeout() builder to both AnthropicClient and
OpenAiCompatClient for per-client timeout configuration
- Parse Retry-After header on 429 responses and use it to override
exponential backoff delay when present
- Add ApiTimeoutConfig to runtime config with apiTimeout settings
in ~/.claw/settings.json (connectTimeout, requestTimeout, maxRetries)
- Add retry_after field to ApiError::Api for propagating rate limit
backoff hints through the retry pipeline
Close two ways the permission system could be bypassed:
- Workspace path traversal: normalize `.`/`..` lexically before the
boundary prefix comparison so paths like `/workspace/../../etc` can no
longer escape the sandbox. Fixed in both the runtime enforcer and the
duplicate check in the tools PowerShell path classifier.
- read-only mode no longer trusts the leading token alone: reject shell
metacharacters (chaining/substitution/redirect/pipe/subshell), drop
interpreters and build drivers (python/node/ruby/cargo/rustc) from the
allow-list, gate `git` to non-mutating subcommands, and reject `find`
actions that execute or delete.
Adds regression tests for both holes. The pre-existing, unrelated
worker_boot git-metadata test failure is not affected by this change.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Commands like /commit, /pr, /issue, /bughunter, /ultraplan are
interactive-only and NOT resume-safe. Previously the generic
interactive_only error always suggested 'claw --resume SESSION.jsonl
/commit', which would just re-trigger interactive_only.
Fix: check commands::resume_supported_slash_commands() in the
SlashCommand::Ok(Some(cmd)) arm. Resume-safe commands get the full
--resume suggestion; non-resume-safe commands only say 'Start claw'.
Also update two existing unit tests whose assertions checked for the old
'interactive-only' substring (now 'interactive_only:' prefix).
Two new integration tests:
- non_resume_safe_interactive_only_hint_omits_resume_suggestion
- resume_safe_interactive_only_hint_includes_resume_suggestion
572 tests pass, 1 pre-existing worker_boot failure unrelated.
/approve, /yes, /deny, /no (and /y, /n) are valid REPL-only slash
commands. Outside the REPL they were falling through to
format_unknown_direct_slash_command -> error_kind:unknown_slash_command.
Fix: intercept them in the SlashCommand::Unknown arm and emit
interactive_only: prefix so classify_error_kind returns the correct kind.
One new test: approve_deny_outside_repl_emits_interactive_only (covers
/approve, /yes, /deny, /no)
572 tests pass, 1 pre-existing worker_boot failure unrelated.
Single-word all-alpha/dash tokens that don't match any known subcommand
now always emit command_not_found (with or without fuzzy suggestions).
Multi-word cases fall through to CliAction::Prompt (natural language
prompt passthrough like 'claw explain this' must still work). The
multi-word gap is documented as ROADMAP #826 (known limitation).
Tests:
- unknown_subcommand_json_emits_command_not_found (new)
- unknown_subcommand_text_emits_command_not_found_on_stderr (new)
- unknown_subcommand_typo_with_suggestions_json_emits_command_not_found (new)
- multi_word_unknown_subcommand_falls_through_to_prompt_826 (documents gap)
572 tests pass, 1 pre-existing worker_boot failure unrelated.
When looks_like_subcommand_typo fires on a single word with no close
fuzzy matches, the fallthrough reached CliAction::Prompt → provider
startup → misleading missing_credentials error.
Fix: always return Err with command_not_found: prefix from the typo
guard (with or without suggestions). Added command_not_found classifier
arm in classify_error_kind. Unified existing unknown_subcommand kind
under command_not_found in #825.
Three new regression tests in output_format_contract.rs:
- unknown_subcommand_json_emits_command_not_found
- unknown_subcommand_text_emits_command_not_found_on_stderr
- unknown_subcommand_typo_with_suggestions_json_emits_command_not_found
Updated pre-existing unit test assertion (starts_with → contains) and
classifier unit test (unknown_subcommand → command_not_found).
572 tests pass, 1 pre-existing worker_boot failure unrelated.
Add SUPPRESS_CONFIG_WARNINGS_STDERR AtomicBool flag in runtime/config.rs
and expose suppress_config_warnings_for_json_mode() via runtime crate.
In main.rs, scan raw argv for --output-format json before parse_args
and activate the flag so no settings-load warnings reach stderr on any
JSON-mode surface (status, sandbox, system-prompt, mcp list, skills list,
agents list, --resume /config*, etc.).
Text-mode surfaces are unaffected; prose deprecation warnings continue
to appear on stderr.
All 572+ tests pass (one pre-existing worker_boot failure unrelated).
* fix: route all JSON-mode abort envelopes to stdout (#819#820#823)
All handled errors in --output-format json mode now write the structured
abort envelope to stdout (rc=1) and keep stderr empty. Previously the
top-level error handler and resume_session JSON branches used eprintln!
which sent the envelope to stderr, breaking machine consumers that read
stdout for command payloads.
Surfaces fixed:
- Top-level abort handler (main.rs): export --session <missing>,
session <subcommand>, prompt (no text), unknown subcommand fallthrough,
flag errors, and all other run() failures
- resume_session JSON branches: session load errors, unsupported commands,
parse errors, command execution errors
Test changes: updated 24 failing contract tests to assert JSON envelopes
on stdout. Added stderr-clean assertions where appropriate. 70 contract
tests pass (was 68; 2 additional from regression coverage).
ROADMAP: #819 (export session-not-found), #820 (interactive_only class),
#823 (missing prompt)
* style: cargo fmt on main.rs after eprintln->println fix
* fix(tests): fmt + update compact_output test for stdout abort envelope routing
* fix(tests): update resume_slash_commands stub test for stdout envelope routing
JSON config output already carries collected config diagnostics in warnings[], so prose stderr emission must be reserved for text/local paths. Lazy permission-mode default resolution prevents an earlier config load from leaking the same deprecation before the JSON renderer runs.\n\nConstraint: ROADMAP #815 requires text mode to keep human stderr warnings while JSON config/list suppresses duplicate app-level config prose.\nRejected: Filtering all stderr in JSON mode | would hide cargo/compiler or unrelated diagnostics outside the app config warning path.\nConfidence: high\nScope-risk: narrow\nDirective: Keep load_collecting_warnings side-effect-free; use load() for human stderr emission.\nTested: cargo fmt; cargo test -p rusty-claude-cli --test output_format_contract config_json_reports_deprecations_structurally_without_stderr_duplicate_815; cargo test -p rusty-claude-cli --test output_format_contract; manual target/debug/claw JSON config fixture.\nNot-tested: cargo clippy -p rusty-claude-cli --all-targets -- -D warnings is blocked by pre-existing runtime dead_code/trident warnings.
Doctor help was already on the local help path in current source, but the exact #702 dogfood surface lacked a focused guard and the JSON help envelope was still too prose-oriented for wrappers. Strengthen the JSON contract while preserving text help.\n\nConstraint: Preserve unrelated dirty rust/Cargo.lock from prior #701 work.\nRejected: Starting runtime/provider/session to inspect doctor semantics | help must be local and credential-free.\nConfidence: high\nScope-risk: narrow\nDirective: Keep doctor help routed through parse_local_help_action and print_help_topic; do not call run_doctor for --help.\nTested: cargo test --manifest-path rust/Cargo.toml -p rusty-claude-cli --test output_format_contract doctor_help -- --nocapture; cargo test --manifest-path rust/Cargo.toml -p rusty-claude-cli --test output_format_contract help -- --nocapture; cargo fmt --manifest-path rust/Cargo.toml --all -- --check; cargo check --manifest-path rust/Cargo.toml -p rusty-claude-cli; timeout 5s cargo run -q --bin claw -- --output-format json doctor --help; timeout 5s cargo run -q --bin claw -- doctor --help.\nNot-tested: full workspace test suite.
Extend auto-compaction error detection to handle additional error patterns
from llama.cpp backends: 'Context size has been exceeded',
'exceed_context_size_error', 'exceeds the available context size'. Also
recover from reqwest 'error decoding response body' errors — some
llama.cpp instances return a non-SSE plaintext HTTP 500 on context overflow,
causing the SSE deserializer to fail.
Add dynamic threshold adaptation: parse server-reported context window
size from error messages (e.g., '(81920 tokens)') and set the auto-
compaction trigger at 70% of that value. This replaces the need for a
hardcoded threshold, adapting automatically to any backend's limits.
This patch was developed with assistance from OpenCode and local Qwen 3.6
API server.
Keep malformed diff invocations with trailing JSON format flags on the parser error path and lock the contract with focused output-format regressions.
Constraint: Do not touch tracked .omx state files.
Rejected: Repeating direct binary smoke loops | local auth/provider configuration intercepts those invocations and obscures parser behavior.
Confidence: high
Scope-risk: narrow
Tested: git diff --check; cargo fmt --check; cargo test -p rusty-claude-cli diff_extra_args_have_typed_error_kind_and_hint_766 --test output_format_contract; cargo test -p rusty-claude-cli diff_trailing_json_after_malformed_args_is_bounded_json_3129 --test output_format_contract; cargo test -p rusty-claude-cli diff_non_git_dir_has_error_kind_and_hint_801 --test output_format_contract
Two classifier arms had no corresponding assert_eq! in
test_classify_error_kind_returns_correct_discriminants: invalid_history_count
(both prefix and contains paths) and unknown_option (#790). Now 49/39 = full
coverage of all classify_error_kind return values.
claw '' and claw ' ' returned empty_prompt + hint:null because the
error message had no newline delimiter. Added usage hint. 61 CLI
contract tests pass.
Parity with #791 (config extra-arg fix). The plugins arg parser emitted
'unexpected extra arguments after claw plugins show ...' with no newline
delimiter, so split_error_hint returned None. Added usage hint after newline.
60 CLI contract tests pass.
bellman
TheArchitectit
Heo, Sung
Sam Lamrabte
YeonGyu-Kim
Bellman
Petter Reinholdtsen