kyle/boxmaker
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: M5c spec: ask_memory at minimal only; Honcho hits must share a term with the query

Browse Source 
From M5c-2 task 00's recordings on rift: a chat at reasoning low returned raw tool-call markup,
and a search with no match still returned the nearest messages.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
kyle
2026-09-28 16:20:57 -07:00
co-authored by Claude Opus 5.5
parent 718e66d3f6
commit 891f41374c
1 changed files with 8 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/specs/2026-09-27-m5c-honcho.md
+8 -3
View File
@@ -199,7 +199,10 @@ The agent's own workspaces are not searched: the local index holds every turn an
agent with exact labels, and would return the same text twice. Each source runs on its own thread agent with exact labels, and would return the same text twice. Each source runs on its own thread
with its own limit (the local index as today; 5 s for each Honcho source); `recall` answers with with its own limit (the local index as today; 5 s for each Honcho source); `recall` answers with
what arrived, and a source that did not answer is one line, `<source>: unavailable`. Hits are what arrived, and a source that did not answer is one line, `<source>: unavailable`. Hits are
ranked within their source and interleaved to `k`, the local index first. Text from Honcho is ranked within their source and interleaved to `k`, the local index first. Honcho's search has no
relevance cutoff (a query with no match still returns the nearest messages, recorded 2026-09-28),
so a Honcho hit is kept only when it shares a search term with the query, as the local index's
full-text match requires. Text from Honcho is
neutralised on arrival and size-capped like every recall snippet. Any `memory-tainted` hit marks neutralised on arrival and size-capped like every recall snippet. Any `memory-tainted` hit marks
the session untrusted, and every hit raises the session's class to its label, as local hits do. the session untrusted, and every hit raises the session's class to its label, as local hits do.
@@ -220,12 +223,14 @@ telemetry, a `secret` session's queries must stop going to Honcho without the ow
## 9. `ask_memory` ## 9. `ask_memory`
A new core tool, `ask_memory { question, workspaces?, reasoning? }`: A new core tool, `ask_memory { question, workspaces? }`:
- `workspaces`: at most two names from `[honcho]`: `own` (default), `untrusted`, `fleet`, or - `workspaces`: at most two names from `[honcho]`: `own` (default), `untrusted`, `fleet`, or
another agent's name from `[[honcho.read]]` (owner, 2026-09-28: two, so that both chats run at another agent's name from `[[honcho.read]]` (owner, 2026-09-28: two, so that both chats run at
once inside `brokerd`'s wait; the model asks again for more); once inside `brokerd`'s wait; the model asks again for more);
- `reasoning`: `minimal` (default: one search, then the answer) or `low`; - Honcho's reasoning level is always `minimal` (one search, then the answer). A `low` chat,
recorded from rift on 2026-09-28, returned the dialectic model's raw tool-call markup as its
answer; the choice comes back when rift answers it;
- `question`: at most 2,000 characters. - `question`: at most 2,000 characters.
One workspace chat (`POST /workspaces/{w}/chat`) per named workspace, both at once; workspace chat searches every conclusion and message of the workspace without the model One workspace chat (`POST /workspaces/{w}/chat`) per named workspace, both at once; workspace chat searches every conclusion and message of the workspace without the model