Imp v0.14.1

Copy Markdown View Source

Imp is a framework for typed, optimizable language-model programs on the BEAM. This release fixes how ReActV2 replays a history that crosses the two finish modes 0.14.0 introduced. Under finish: :submit, a plain-text answer recorded under finish: :auto is no longer taken for an unsent thought. Under finish: :auto, a thought-only step recorded under finish: :submit is followed by "Your message above was not sent." instead of standing as a sent message. It adds no options and changes no public API. A history written only in a plain-text loop replays byte-for-byte as in 0.14.0, and under finish: :submit only steps that carry the signature's outputs are treated differently. These are fixes to the opt-in finish: :submit that 0.14.0 added, so under Imp's versioning rule this is a patch release.

Install

{:imp, "~> 0.14.0"}

~> 0.14.0 takes later 0.14 patch releases and nothing newer. The two-part form, ~> 0.14, means >= 0.14.0 and < 1.0.0 in Mix, so mix deps.update imp under it takes every later 0.x release, including minor releases that change what callers receive.

Every dependency comes from Hex. Use a path dependency only while developing against a local checkout.

ExMCP and erlexec are declared runtime: false, so an OTP release that uses Imp.ACP or Imp.MCP must list applications: [ex_mcp: :load, erlexec: :load] in its release definition; see releases that use MCP or ACP. Ordinary Imp startup starts no protocol endpoint.

Changes and upgrading from 0.14.0

Under {:imp, "~> 0.14.0"}, run mix deps.update imp and commit mix.lock. No code or data migration is needed.

  • Fixed: an answer step is never a thought. Under finish: :submit, a step recorded under finish: :auto as the turn's plain-text answer (text and the signature's outputs both) was treated as a thought-only step when its history was passed on: the "Your message above was not sent." line followed each answer that had been delivered, and models sent those answers again. A step that carries the signature's outputs is an answer: it gets no line, and the forced submit never moves it to the user side. Only a step with text, no calls and no outputs is a thought.
  • Fixed: a plain-text loop keeps roles alternating. In a plain-text ReActV2 loop (finish: :auto), a history written under finish: :submit replayed each thought-only step as an assistant message, which there reads as a message that was sent, often right before the answer's: two assistant messages in a row. Each such step is now followed by a user message, "Your message above was not sent." (the line's factual half; a plain-text loop has no submit to name). A history written only in a plain-text loop has no thought-only step, so its requests are byte-for-byte unchanged.
  • Known gap. Under finish: :submit, a forced request with nothing new to ask, on a history that ends with a plain-text answer step, still ends on that assistant message, since an answer is not moved. Resume with new inputs or a last_request_note.

Rollback means restoring 0.14.0 in the lock; mixed-mode histories then replay as described above before each fix. See the changelog for every change.

Known limits

  • The usage map (:usage on the :model_response event, Imp.Usage, Imp.Prediction.get_lm_usage/1) is ReqLLM's, with input_includes_cached added. Its :cost and :total_cost are ReqLLM's catalog estimate, not a charge, for streamed and non-streamed calls alike, and OpenRouter's charge is its "cost". Read money from cost and estimated_cost.
  • An error the provider sends inside a stream has status nil and retryable: true, because ReqLLM's stream decoder keeps only its message; the same error in a non-streamed response may carry a status that says not to retry.
  • A flat two-element name-first list that is not itself an element of a list (at the top level, in a tuple, or under a key that is not a credential name), such as ["password", "hunter2"], is not redacted, where 0.5.0 redacted it. Such a list is read as a key and value only as an element of a list, so that lists of names such as with_inputs([:api_key, :question]) survive saving.
  • The GEPA, SIMBA, MIPROv2, InferRules and random-search checkpoints redact their failure reasons but otherwise hold the resume state as it is: GEPA's candidates, evaluation cache, proposed instructions and reflection data, and the others' instructions, demos and scores. Fast-Slow and Playbook checkpoints likewise hold their prompts, rollouts and playbooks. Treat checkpoint files as sensitive.
  • A ReActV2 step whose prose quotes its own field names (a JSON object with next_thought or tool_calls keys, a [[ ## field ## ]] line, a <next_thought> tag) is not read as prose and goes to the JSON fallback, which usually costs one more call. If the model repeats the same reply, the step ends :incomplete. Beside native tool calls, such a thought becomes nil; the calls still run.
  • Tool images reach the model as user messages after the turn's tool responses, labelled with the call, not nested in the tool result, even for providers that accept images in tool results. Only MCP text and image blocks are converted; other block types stay text, image URLs in text are never fetched, and the selected model must accept image input of that format and size.
  • ATIF's extra.seen_by_model and the read-as content come from matching a tool message in the next captured request. A result that request does not carry, including one that ended the run, keeps the tool's own output and is marked unseen. agent.tool_definitions comes only from :tools_sent events, and final_metrics totals are left out when any response lacks that figure.
  • A streamed completion that the provider's content filter stops still ends as a completion: the stream's finish_reason is :content_filter and the filter's text has already arrived as chunks. Only non-streamed requests turn it into an Imp.LMError.
  • When a streamed ReqLLM request fails, ReqLLM's own warning log prints the error with its response headers. Imp strips them from the error it returns, but cannot change that log line.
  • In command output, a Bearer token of 12 to 15 characters, or one with no digit, is shown when more text or a newline follows it. 0.5.0's command pipeline hid any such token of 12 or more characters.
  • A GEPA checkpoint taken between preparing and starting a full validation resumes by rejecting that candidate, as before.

The CHANGELOG records every user-visible change in this release. Generated module documentation is the complete API reference. Start with Imp, Imp.Signature, Imp.Module, Imp.Evaluate, Imp.Optimizer, Imp.ACP, Imp.MCP, and Imp.Telemetry.