fix: buffer exec output for next user turn; alternation for strict templates

User-test surfaced the bug: with `deep` (mistral-nemo-12b) active,
running `list files` -> y on `CMD: ls` -> `Are there directory entries
beginning with "lor"?` returned a Jinja exception:

    api: ... Error: Jinja Exception: After the optional system message,
    conversation roles must alternate user/assistant/user/assistant/...

Cause: §6 specified "exec output injected into context uses role 'user'
with a prefix tag '[exec output]'." This works for permissive templates
(qwen2.5-coder-1.5b, the `fast` preset) but produces a back-to-back
user/user pair on strict templates that enforce the OpenAI alternation
contract — `[exec output]` user turn followed by the user's actual
follow-up question.

Fix:

context.lua:
  - new field `pending_exec_output` (initially nil)
  - new method `:append_exec_output(out)` buffers (concat on subsequent
    captures so multi-shell-then-ai still merges everything)
  - new method `:append_user(content)` flushes buffered exec output as
    a `[exec output]\n...\n\n` prefix and appends a user turn
  - `:reset()` also clears the buffer

repl.lua:
  - run_shell calls ctx:append_exec_output(out) instead of
    ctx:append({role="user", content="[exec output]\n"..out})
  - ask_ai calls ctx:append_user(text) instead of raw :append; saves
    prev_pending so a broker error can restore the buffer for retry

PHASE0.md §6:
  - amended the role-injection paragraph to describe the buffer-and-
    prepend policy; the §3 invariants list is untouched (this was a §6
    design detail, not a locked invariant)

Verification:
  - context unit tests cover: alternation after the failing sequence,
    multi-shell merge, reset clears buffer, broker-error retry path
  - live reproduction against `deep` (mistral-nemo) of the exact
    user-reported sequence succeeds; model responds with a sensible
    `CMD: ls | grep '^lor'` instead of a Jinja exception

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-10 18:41:21 +00:00
parent 8870eb0451
commit 16490e6905
3 changed files with 42 additions and 10 deletions
+30 -4
View File
@@ -20,10 +20,11 @@ Context.__index = Context
function M.new(opts)
opts = opts or {}
return setmetatable({
system_prompt = opts.system_prompt or DEFAULT_SYSTEM_PROMPT,
turns = {},
max_turns = opts.max_turns or 40,
token_budget = opts.token_budget or 4096,
system_prompt = opts.system_prompt or DEFAULT_SYSTEM_PROMPT,
turns = {},
pending_exec_output = nil, -- buffered until next user turn (§6)
max_turns = opts.max_turns or 40,
token_budget = opts.token_budget or 4096,
}, Context)
end
@@ -33,6 +34,30 @@ function Context:append(turn)
self.turns[#self.turns + 1] = { role = turn.role, content = turn.content }
end
-- Buffer captured shell-exec output. Per §6 (post user-test fix), exec output
-- is NOT appended as its own user turn — strict chat templates (e.g. mistral-
-- nemo's Jinja) reject the resulting user/user back-to-back. Instead it is
-- held until the next user turn arrives, then prepended via :append_user.
function Context:append_exec_output(out)
if not out or out == "" then return end
local block = "[exec output]\n" .. out
if self.pending_exec_output then
self.pending_exec_output = self.pending_exec_output .. "\n" .. block
else
self.pending_exec_output = block
end
end
-- Append a user turn, flushing any pending exec output as a prefix. Use this
-- (rather than raw :append) for any turn whose role is "user".
function Context:append_user(content)
if self.pending_exec_output then
content = self.pending_exec_output .. "\n\n" .. content
self.pending_exec_output = nil
end
self:append({ role = "user", content = content })
end
-- Render the messages array for broker.chat (system prompt prepended; turns
-- in order). The system prompt is NOT stored in self.turns per §6.
function Context:to_messages()
@@ -72,6 +97,7 @@ end
function Context:reset()
self.turns = {}
self.pending_exec_output = nil
end
return M