wechat-cli v1.6.19

wechat-cli v1.6.19

• Refines tool effect metadata without breaking the existing local_write boolean.
• Adds local_write_mode: none / possible / required.
• Adds strict_read_only_behavior: same / allowed_without_writes / blocked.
• Marks search as none/same, timeline/search-context/tail/media as possible/allowed_without_writes, and export/cache refresh/rebuild as required/blocked.
• Adds tests and keeps release package defaults at 1.6.19.

wechat-cli v1.6.18

wechat-cli v1.6.18

• Adds read_only/local_write effect metadata to tool schemas so agents can distinguish WeChat read-only operations from local support/output writes.
• Makes status degradation self-explanatory with live_read_ok and degraded_by fields.
• Removes misleading before_count/after_count query echo when search-context runs with context_limit=0.
• Clarifies call/call-json documentation to prefer command-or-tool aliases and documents stats as metadata-cache stats, not message/search trend stats.
• Keeps pattern-search/search-stats out of the base CLI layer for now.

wechat-cli v1.6.17

wechat-cli v1.6.17

• Lets call and call-json accept command names and aliases, not just raw tool names, so search-context/tail/watch resolve to their canonical tools.
• Tightens search-context schema and runtime validation for limit, context_limit, before_count, after_count, and aliases.
• Makes search-context before_count/after_count default to 5 independently of search limit.
• Echoes context_limit=0 and contexts_returned=0 when no context expansion is explicitly requested.
• Adds regression tests for callable alias resolution, search-context bounds, and context window defaults.

wechat-cli v1.6.16

wechat-cli v1.6.16

• Treats sender=me/self as from_me=true in query metadata across agent-facing read outputs.
• Adds include_text, max_text_chars, and snippet_only to search-context for both hit rows and expanded context messages.
• Adds text_truncated/original_text_chars metadata to trimmed search-context context rows.
• Adds schema bounds for tail/read_events limit and scan_limit.
• Bumps release package defaults to 1.6.16 and keeps MCP files out of release zips.

wechat-cli v1.6.15

wechat-cli v1.6.15

• Keeps status thin but less misleading: metadata-cache gaps now degrade readiness instead of blocking live reads, and status exposes concise capability booleans.
• Adds schema minimum/maximum bounds for search limit, offset, and max_text_chars.
• Echoes from_me in query metadata for search/timeline/search-context/tail message mode.
• Adds text_truncated and original_text_chars to truncated/omitted search rows.
• Adds tests for capability splitting, schema bounds, query echo, and search truncation metadata.

wechat-cli v1.6.14

wechat-cli v1.6.14

• Adds --version / version as a script-friendly CLI entrypoint.
• Makes root help command objects expose a stable name field while retaining command.
• Adds sender=me / sender=self and from_me=true across agent read surfaces that filter senders.
• Fixes search pagination by honoring offset in the live FTS path.
• Adds search light-output controls: max_text_chars, snippet_only, and include_text=false.
• Tightens search limit validation and adds tests for tail JSONL/cursors, forwarded unresolved voice/file warnings, regression script inputs, and MCP help/package regressions.
• Removes remaining public WX_MCP_* fallback documentation from release/install/package surfaces while preserving legacy cleanup paths.

wechat-cli v1.6.13

wechat-cli v1.6.13

• CLI-only release: removed MCP runtime, manifest, installer registration, and release packaging surface.
• Added agent-first CLI entrypoints: agent, status, coverage, workflows, context, search-context, and tail / watch.
• Added strict local read-only mode via --strict-read-only / WECHAT_CLI_STRICT_READ_ONLY=1.
• Slimmed default agent schemas; use tool-schema <cmd> --profile all or tools --profile all for compatibility/debug fields.
• Added real WeChat read regression script for release validation.

wechat-cli v1.6.12

Fix macOS release installs that can leave wechat-cli killed at startup after an update.

• Install macOS binaries via temp file, ad-hoc codesign, then atomic rename.
• Avoid in-place overwriting the running Mach-O during update.
• Document the recovery path for zsh: killed ~/.local/bin/wechat-cli --help.

For a broken macOS install that cannot run wechat-cli update, recover with:

curl -fsSL https://raw.githubusercontent.com/r266-tech/wechat-cli/main/scripts/install-release.sh | zsh -s -- --update

wechat-cli v1.6.11

Release 1.6.11 fixes macOS wxkey PBKDF fallback target handoff for WeChat 4.1.10+.

Changes:

• PBKDF fallback now stops existing WeChat before launching the debugged instance.
• When the original WeChat app is already ad-hoc signed, wxkey launches the original app under LLDB instead of always launching the shadow copy.
• This addresses failures with diagnostics like breakpoint_stops=0 / pbkdf_calls=0 when original WeChat kept the login/DB state while the shadow process did not decrypt DBs.
• Docs and installer next_action now describe the new fallback handoff behavior.

Verification:

• wxkey: go test ./...
• wechat-cli: go test ./...
• zsh -n install.sh
• zsh -n scripts/package.sh
• zsh -n scripts/install-release.sh
• macOS zip checksum verified
• Windows Release Package workflow 26926746143 passed, including installer smoke

wechat-cli v1.6.10

Release 1.6.10 improves macOS wxkey no-key diagnostics and first-run fallback behavior.

Changes:

• Bundle wxkey diagnostics that distinguish no PBKDF calls, wrong DB root/account, and matched-salt-but-unverified-key failures.
• Make PBKDF fallback quality-first with a 5-minute default probe.
• Rename scan progress "verified" wording to "candidate_checks" in wxkey output.
• Update README, llms.txt, AGENTS.md, and installer next_action guidance for PBKDF no-key failures.

Verification:

• go test ./...
• zsh -n install.sh
• zsh -n scripts/package.sh
• zsh -n scripts/install-release.sh
• macOS zip checksum verified
• Windows Release Package workflow 26874287580 passed, including installer smoke