Skip to content

merge dev into master#454

Open
Tarquinen wants to merge 25 commits intomasterfrom
dev
Open

merge dev into master#454
Tarquinen wants to merge 25 commits intomasterfrom
dev

Conversation

@Tarquinen
Copy link
Collaborator

@Tarquinen Tarquinen commented Mar 24, 2026 •
edited
Loading

Closes #450
Closes #452

  • Add `compress.mode` config setting, defaults to `"range"` (previous behavior) and new experimental `"message"` option available.
  • `"message"` mode allows DCP to more surgically compress individual messages in a batch rather than a full range.
  • `"message"` mode allows DCP to determine compression priority for each message based on how many tokens the message is (low/med/high for 0-500/500-5000/5000+ tokens), not configurable yet but maybe in the future?
  • Nudges remind the model the IDs of the high priority messages in `"message"` mode to ensure they don't get left behind.
  • `"range"` mode now supports the model using a list of ranges instead of just 1 range if it wants.
  • `compress.flatSchema` deprecated, not needed for lower tier model performance anymore due to new tool schemas (I think...)
  • Bug fixes and prompt rearrangements and code cleanup refactoring

Tarquinen and others added 25 commits March 17, 2026 01:57
@Tarquinen
compress: add message mode
b75aad8
@Tarquinen
simplify compress mode handling
baf6e1d
@Tarquinen
remove message arg normalization
111ca2a
@Tarquinen
simplify compress notifications
8d04298
@Tarquinen
sync config docs and schema
acb6f51
@Tarquinen
remove stale compress mode text
08c7fcd
@Tarquinen
batch compress tool inputs
9d1f053
@Tarquinen
clean up compress config leftovers
d480b17
@Tarquinen
refactor compress module layout
7e850dc
@Tarquinen
group compression runs by tool call
7116280
@Tarquinen
document experimental message mode
9c8aa18
@Tarquinen
prioritize high-cost message compression
41d1aea
@Tarquinen
add message token count inspector
6ddb0f2
@Tarquinen
separate prompt responsibilities between system and tool specs
fcfebbe
@Tarquinen
append injections to text parts
3da14ff
@Tarquinen
mark blocked refs in message mode
aa4f76a
@Tarquinen
prefer tool output for assistant ids
9ca33e4
@Tarquinen
clean up unused compression types
5a13480
@Tarquinen
Merge pull request #446 from Opencode-DCP/feat/compress-message-mode
5d5b8c2 
compress: add message mode
@Tarquinen
fix stale tokens after /compact (#450)
90d4089 
Treat the native compaction summary as stale token data so context nudges wait for the first fresh assistant response.
@Tarquinen
reset message ids after /compact
425339f 
Clear DCP message alias state when native compaction is detected so refs rebuild from the compacted conversation instead of continuing toward m9999.
@Tarquinen
inject ids into all parts
592a5dd 
Keep message references visible on every text chunk and completed tool output so multi-part messages stay fully addressable.
@Tarquinen
prioritize compress tool messages
2dc15e1 
Keep completed compress calls visible as high-priority message targets so message-mode guidance preserves them consistently.
@Tarquinen
count all tool inputs
ee9ac10 
Match message token estimates to the tool arguments OpenCode stores so large built-in, custom, and plugin tool inputs do not slip past sizing.
@Tarquinen
refactor isIgnoredUserMessage contract
8562ccc 
Move the user-role guard into the helper so callers can use it consistently across message flows. Add a regression test to keep non-user messages from being treated as ignored.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

1 participant

@Tarquinen