dimagi-internal/ace / skills/llo-feedback
llo-feedback
UnverifiedAsk an opportunity's LLOs for feedback on the application, process, and next steps, and record their responses.
LLO Feedback
Collect feedback from LLOs about the completed opportunity.
Inputs
| Parameter | Used for | Resolver |
|---|---|---|
opportunity_id, organization_slug | the opportunity the feedback is about; delivery metrics | explicit arg; default: run_state phases.solicitation-management.products.selected_llo.opportunity_id + the run's holding org; else the program context doc |
llo_contacts[] | who is asked (name, email, org) | explicit arg; default: run_state selected_llo + the run's comms-log; else the program context doc |
archetype | which question set (§ Archetypes) | explicit arg; default: PDD / launch record frontmatter; else atomic-visit |
run_id | run-bound mode: comms-log + 10-closeout/ record | optional; absent = live-program mode |
out_folder_id | where the feedback record goes in live-program mode | explicit arg when run_id is absent |
Live-program mode (no run_id): metrics come live from
connect_get_deliver_progress on opportunity_id; message ids are logged to
the program project's board task (or the context doc) instead of a run
comms-log, and Step 6 writes to out_folder_id. Sends stay approval-gated in
both modes (external LLO contact).
Process
-
Read opportunity context from GDrive:
- LLO contact info from invite/comms logs
- Opportunity summary (including archetype from PDD or launch record)
- Key metrics — archetype-specific (delivery rates for atomic-visit; session completion + audio-quality stats for focus-group; stage completion + transition-gate outcomes for multi-stage)
-
Read the
archetype:field. The feedback questions — what "usability" means, what "FLW experience" means, what improvement areas are worth asking about — all shift by archetype. See## Archetypesbelow. Fall back toatomic-visitif unspecified. -
Compose feedback request email (archetype-aware):
- From: ACE's mailbox (
config/agent.jsonemail; sent via theemail-communicatorskill, never a shell var) - To: each LLO contact email
- CC: CRISPR Admin Dimagi Google Group
- Questions branch per archetype — see
## Archetypes. Keep the core envelope consistent (support experience, suggestions, future interest) across all archetypes.
- From: ACE's mailbox (
-
Send the feedback request emails via the
email-communicatorskill (or draft for review). Log message IDs to the opp's comms log for response correlation. -
Monitor for responses via OCS transcripts or email.
-
Document feedback to
ACE/<opp-name>/runs/<run-id>/10-closeout/llo-feedback.md:- Archetype recorded explicitly so
learnings-summaryandcycle-gradeknow how to aggregate - Responses from each LLO
- Common themes
- Specific improvement suggestions (tagged by archetype dimension: app usability / session facilitation / stage transitions / support / training / other)
- Archetype recorded explicitly so
Archetypes
Feedback questions that miss the work the LLO actually did produce thin responses and training data that drifts toward whatever archetype was front-loaded in the question. Ask about "app usability" when the LLO ran focus groups, and you get a shrug plus nothing learnable about the facilitation that actually happened.
atomic-visit (default)
Feedback questions:
- App usability: Learn app (training modules clear? friction?), Deliver app (forms, case management, offline behavior)
- FLW experience: what did FLWs find hard? what tripped them up? what did they flag as buggy vs. confusing vs. just hard?
- Field conditions: signal / battery / language / literacy issues? anything the app assumed that wasn't true?
- Process experience: onboarding, support responsiveness, communication cadence
- Suggestions for improvement: open-ended
- Future interest: willingness to participate in similar opps
focus-group
Feedback questions (replaces app-usability block — the LLO has no app):
- Question guide: which questions worked? which questions produced confused / thin / deflected answers? was the sequencing natural?
- Facilitation experience: venue, recording equipment, participant comfort, consent flow. Any moments where the technology got in the way?
- Audio + upload workflow: file sizes, upload reliability, did the session-note template capture what mattered? what was redundant vs. missing?
- Participant recruitment: did the PDD's eligibility criteria produce representative participants? any segments under- or over-represented?
- Session cadence: was the session length right? did the N-sessions schedule land well with participants' availability? what about the 48h write-up deadline?
- Process experience: onboarding, support responsiveness, communication cadence (shared across archetypes)
- Suggestions for improvement: open-ended, with a specific prompt for "what would make the question guide better for a follow-up round?"
- Future interest: willingness to facilitate similar pilots
Why the swap matters: the atomic-visit questions don't just waste the LLO's time — they signal to the facilitator that ACE doesn't understand what they did, which colors the whole response. A facilitator asked "how was the Learn app?" after running discussion groups hears "we weren't paying attention" and writes shorter answers to the rest of the survey.
multi-stage
Feedback questions: split per-stage. For each stage the LLO owned, apply the matching archetype's question set. Cross-stage additions:
- Stage transitions: was the hand-off between stages clear? did you have what you needed from the prior stage when your stage started? did the stage-transition gate give you enough time?
- Pipeline coherence: looking across all stages, did the work feel like one opportunity or N disconnected ones? any context you expected to carry through the pipeline but didn't?
Feedback is typically collected once per LLO, not per-stage, to avoid survey fatigue. If an LLO owned 3 stages, their feedback covers all 3 in a single response.
MCP Tools Used
- Google Drive:
drive_read_file,drive_create_file - OCS:
ocs_list_sessionsfor monitoring responses - Email:
email-communicatorskill (sends fromace@dimagi-ai.com)
Mode Behavior
- Auto: Send feedback requests, monitor and document responses
- Review: Present email drafts for review before sending
Dry-Run Behavior
When --dry-run is active:
- Write feedback request email drafts (recipients, subject, body) to
comms-log/dry-run-llo-feedback.md - Do not send emails or monitor for responses
- State tracks as
dry-run-success
Terminology
The platform is Connect, never CommCare Connect — in every word this skill
puts in front of a human. The full rule, including the carve-outs for CommCare the
mobile app, CommCare HQ and code-level identifiers, is
skills/_terminology.md. Read it before writing prose; it is
binding on this skill's output.
Change Log
| Date | Change | Author |
|---|---|---|
| 2026-10-18 | Add ## Inputs as parameters (opportunity, LLO contacts, archetype) with run defaults, plus live-program mode; sender is ACE's mailbox, not the retired $ACE_GMAIL_ACCOUNT (ace#1147). ace#3148. | ACE (build-vs-maintain audit) |
| 2026-04-20 | Added ## Archetypes with per-archetype feedback questions. focus-group replaces the app-usability / FLW-experience block with question-guide quality, facilitation experience, audio+upload workflow, participant recruitment, and session cadence. multi-stage asks per-stage questions plus cross-stage transition quality. Prevents the "facilitator asked about a Learn app they never used" anti-pattern that drifts responses thin | ACE team |
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 192k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 192k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 192k | Updated | Claude CodeCodex | Verified | Install |