feat(im): make IM operations agent-reliable - #2106

Open

luozhixiong01 wants to merge 41 commits into

mainfrom

feat/im-agent-reliability-closeout

Commits on Jul 31, 2026

feat(im): add copyable Tips examples to high-frequency IM shortcuts
luozhixiong01
committed
feat(im): add raw API affordance guidance for the im domain
luozhixiong01
committed
fix(im): seed failure inventory with recovery hint fix and coverage updates
luozhixiong01
committed
docs(im): relax send confirmation gate for fully specified requests
luozhixiong01
committed
feat(im): extend Tips examples to feed and flag shortcuts
luozhixiong01
committed
fix(im): add id-source hint and strengthen example and recovery locks
luozhixiong01
committed
docs(im): require draft approval when message content is delegated
luozhixiong01
committed
docs(im): surface sending approval semantics in the domain skill
luozhixiong01
committed
docs(im): cover reply target in domain approval rule and tighten hint lock
luozhixiong01
committed
docs(im): forbid downgrading reply intent to a new direct message
luozhixiong01
committed
test(im): drop redundant loop variable copy and type error assertions
luozhixiong01
committed
docs(im): mention the user-id direct message form for plain text sends
luozhixiong01
committed
fix(im): fail closed on media upload errors instead of rewriting content
luozhixiong01
committed
docs(im): extend approval semantics to all outbound actions and card drafts
luozhixiong01
committed
test(im): lock example identity, run all examples, guard affordance drift
luozhixiong01
committed
test(im): route config secret through printf placeholder for content scan
luozhixiong01
committed
test(im): use the scanner-recognized test-secret placeholder inline
luozhixiong01
committed
fix(im): pin explicit identity on user-only examples and run them verbatim
luozhixiong01
committed
docs(im): pin explicit identity on every outbound example command
luozhixiong01
committed
docs(im): pin bot identity on image pre-upload examples
luozhixiong01
committed
feat(im): enforce trustworthy completion for IM writes
luozhixiong01
committed
feat(im): expose trustworthy completeness for IM reads
luozhixiong01
committed
chore(im): align help skills and contract coverage
luozhixiong01
committed
fix(im): keep contract exemptions public-safe
luozhixiong01
committed
refactor(im): make acceptance-only a first-class contract
luozhixiong01
committed
fix(im): close completion contract gaps
luozhixiong01
committed
feat(im): require idempotent chat creation
luozhixiong01
committed
feat(im): add structured message mentions
luozhixiong01
committed
fix(im): close read completeness gaps
luozhixiong01
committed
chore(im): align agent guidance and coverage
luozhixiong01
committed
fix(im): make agent guidance task-scoped
luozhixiong01
committed
docs(im): state conservative mention evidence
luozhixiong01
committed
test(im): align acceptance contract coverage
luozhixiong01
committed
fix(im): close completion contract review gaps
luozhixiong01
committed
fix(im): expose defaulted identity and align help layout
luozhixiong01
committed
fix(im): refine identity and mention help signals
luozhixiong01
committed
fix(im): remove superseded helpers
luozhixiong01
committed
chore(im): trim review-only artifacts
luozhixiong01
committed
test(im): use approved fixture domain
luozhixiong01
committed

Commits on Aug 1, 2026