Conversation
This reverts commit 96ad4a5.
This was implemented in main and this commit is syncing the telgram implementations of last_message with main
Implemented telegram security/capability matrix
Also made the function configurable through CLI input
Load BOT_TOKEN/CHAT_ID from .env via dotenv instead of passing through metta config. Refactor channel selection from case to if chains. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add prompt-configure that asks for each setting interactively with defaults shown. CLI args (name=value) still override and skip the prompt. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Added frame composition and frame management skills
…-parent frame relation id
…rameRef indexspace removal - cfv2-load-frame: add sread normalisation guard as first let* binding to strip quotes from Grounded frame ID atoms passed by the LLM (e.g. Frame-...) so cfv2-get-frame pattern match no longer silently returns () - cfv2-make-frame-ref-from-current: fix status field destructuring to use (status ) so the named field is correctly re-wrapped in FrameRef output; fix debug println to reference instead of ' - add cfv2-check-ref-id and cfv2-remove-frame-ref to correctly match and remove FrameRef atoms from frame-indexspace (cfv2-remove-frame only matches Frame atoms and was silently failing on the indexspace) - use cfv2-remove-frame-ref in cfv2-complete-current-frame-to-stm/ltm to correctly clean up frame-indexspace on frame completion
Replace flat 2-branch signal with 4 distinct cases driven by and frame state: - Case 1: new input, no prior active frame — clean start signal - Case 2: new input, prior frame was active — switch-frame hint signal - Case 3: no new input, frame loaded — continue or idle signal - Case 4: no new input, no frame — autonomous goal hint signal Add binding captured before ctx-ingest-user-message to correctly distinguish Case 1 from Case 2 — after ingestion a new frame is always present making the two cases indistinguishable at prompt time. Legacy useFrames=False path preserved with original DO NOT RE-SEND OR SPAM! signal.
- replace cfv2-check-ref-id and cfv2-remove-frame-ref with a second cfv2-check-frame-id clause that pattern-matches FrameRef, allowing cfv2-remove-frame to handle both Frame and FrameRef spaces uniformly - update cfv2-remove-frame comment to reflect it now dispatches on both Frame and FrameRef via cfv2-check-frame-id - fix swapped case comments in loop.metta block to match actual branch order of the if expression - remove unnecessary comment above binding
…rame fix(context): frame ID normalisation, FrameRef fixes and 4-case loop signal
…ew modifications on the context frame skills description
With context frames the received message is no longer passed after the ":-:-:-:" delimiter, so the mock never found a registered answer and every test that drives the agent failed. - match the request against the CURRENT_CONTEXT_FRAME_S_EXPR section when the delimiter carries a loop signal, keeping the HUMAN-MSG path first so the suite still runs on core main - serve an answer once per frame and complete the frame afterwards, so an unfinished frame does not keep later messages from becoming current - let a test keep the frame open with set_answer(..., complete_frame=False) - drop registered answers between tests
… default frame embedding model.
Disable Chroma's default embedding function because frame relations provide their own embeddings, and skip similarity queries when no other frames exist. Remove duplicate frame-completion operations from skills.metta so the canonical implementations in context.metta are invoked only once.
Frame completion mutates memory and frame spaces but stays a nondeterministic call, while the command dispatcher evaluates commands under collapse. Every retry re-executed the whole chain with its side effects, each pass produced a different result, and the agent never returned. Wrap the three completion commands in once. balance_parentheses split an answer by lines and then stripped the outer parens of the whole block. An answer whose first form takes no arguments was corrupted: the command name kept its closing paren and the remaining form became a string, so the agent reported a syntax error and ran none of the commands. Split a block into top-level forms first, ignoring parens inside string literals.
…-mock [OMEGA-286] Match mock answers against the context frame
…-completion-loop [OMEGA-286] Fix frame completion loop and multi-form answer parsing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
When the configured LLM provider's API quota or billing limit is reached, the agent was silently failing, returning empty responses every loop iteration with no feedback to the user. This fix detects quota/billing exhaustion errors across all supported LLM providers and sends a one-time notification to the user through the active communication channel (Telegram, Slack, IRC, Mattermost, etc.).
Changes made:
lib_llm_ext.py to detect quota errors from any provider using known error signal strings