Competitor Monitoring
You watch competitors for a salesperson. They read the result in an app, not
in the chat — the Sales Companion iOS app parses a fenced watch block from
your daily memory note. Your job is not to write a nice report but to detect
reliably what changed since the last run, and to leave out everything else.
Keep the tone factual. No "exciting", no "interestingly". A salesperson reads
this in twenty seconds in the morning. Write names, titles and details in the
user's language.
The four stores
| What |
Where |
How |
| Schedule |
one cron task per target |
cron add / update / remove — chat only |
| Watchlist |
watch/targets.json in the workspace |
read / write |
| Comparison state |
watch/<target-id>.json in the workspace |
read / write |
| Result for the app |
today's daily note |
memory with action: "append", target: "daily" |
watch/targets.json is the watchlist. Scheduled runs cannot call cron
at all — the tool is blocked there — so they can only know the full list from
this file. Change it in the same step as the cron task, every time:
{"targets": [
{"id": "<target-id>", "name": "<display name>", "kind": "company",
"focus": "<what to watch>", "schedule": "<e.g. täglich 07:30>",
"url": "<website>", "task_id": <number from cron add>}
]}
This and every other <…> shape in this file is a template: fill in
real values from the user's request, watch/targets.json or this run. Never
write a template, or an example company name from this file, into any file.
task_id is the number cron add reports ("Scheduled … task #42 …"); pass
it as taskId to update and remove.
Why these stores: the memory tool may only write today's daily note, and
only memory files are synced to the platform, where the app reads them. A
workspace file never reaches the app, so it holds your own state.
A. Set up a watch
This happens in a chat with the user, where cron is available.
Clarify who and what to look for. If the focus is missing, ask
exactly once ("What should I watch at Allianz — prices, products,
people?"). If the website is missing, look it up yourself; do not ask.
Derive a stable id from the name: lowercase ASCII, hyphens, no
umlauts — allianz-de, barmenia, jane-doe. If watch/targets.json
already has it, update that watch instead of adding a second one.
Create the cron task:
cron action=add
cron="<minute> <hour> * * *"
tz="<user's IANA time zone>"
channel="<the user's email address>"
prompt="Competitor monitoring: use the competitor-monitoring skill,
read its SKILL.md and follow it. Target: <target-id>
(<display name>, <website>), kind=<company|person>,
focus: <what to watch>."
- Pass
tz with the user's IANA time zone so the run keeps its local time
across daylight-saving changes. If you do not know it, ask; for a
German-speaking user, Europe/Berlin is the usual answer.
channel is required when the request comes from a web chat
session: the tool refuses to schedule there, because the output of a task
created without a delivery channel would be discarded. Use the user's
email address or another configured messaging target. If none is known,
say so plainly instead of trying without one.
Add the target with its task_id to watch/targets.json (create the file
if it is missing).
Do not create watch/<id>.json. The first scheduled run records the
baseline; a file created now would only be empty.
Write the result block right away (section C) with the updated list and
"findings": []. Otherwise the new target only appears in the app after
the first scheduled run.
Confirm in one sentence: "I'm watching Allianz from tomorrow morning,
focus pet insurance."
Changing a watch (time, focus): cron action=update taskId=<task_id>
from watch/targets.json, then update the entry there. Never add a second
task for the same target.
Removing a watch: cron action=remove taskId=<task_id>, remove the
entry from watch/targets.json, delete watch/<id>.json, write the block
again.
B. A scheduled run
You receive the task's prompt. cron is not available here; do not try it.
Work strictly in this order:
Read the watchlist and the previous state. read watch/targets.json
and read watch/<id>.json. If the state file is missing or its
snapshots are empty, this is a baseline run: fetch the sources, save
them as the state (step 6) and report nothing as a finding — there is
nothing to compare against yet. Still write the block with
"findings": [].
Fetch the current state. Use several sources, not just one:
web_search with freshness: "week" on company name plus focus
web_fetch on the pages that belong to the focus (pricing, product,
press pages) — they are listed in the state under snapshots
- for
kind=person, also search name plus role. Social activity is only
partly visible with the available tools, so rely on interviews, talks,
press releases and job changes.
Compare. For each source: did the substance change? Ignore counters,
dates, cookie banners, ordering and rewording that does not change meaning.
Judge. A finding only counts if it falls into one of these categories.
Use the category names as given; the app shows them as labels.
category |
Example |
Preis (price) |
plan more expensive, new discount, tiers changed |
Produkt (product) |
new product, benefit dropped, terms changed |
Partnerschaft (partnership) |
cooperation, reseller, integration |
Finanzierung (funding) |
funding round, acquisition, sale |
Personal (people) |
change in management or sales leadership |
Stellen (hiring) |
notable job ads that reveal a direction |
Everything else — guide articles, blog posts, social chatter, redesigns,
anniversaries — is severity: "minor" or does not belong in the block at
all. severity: "significant" is only for what the salesperson must know
today because it changes a conversation; only significant triggers a
notification on their phone. When in doubt, minor: a phone that buzzes
too often gets muted.
Deduplicate. Give every finding a fingerprint: category, source URL
and the new value in a few words — Preis|https://allianz.de/preise|Premium 27,90 €. If the state's reported map already has that fingerprint, the
change was reported before: drop it from this run. Otherwise assign a new
id (see section C).
Write the result block first (section C). Only if the memory append
succeeded, go on to step 7. If it fails — for example because today's note
is full — stop and leave the state untouched, so the next run finds the
same changes again instead of losing them. Reply with the failure line
(step 8).
Then update the state. write watch/<id>.json:
{"target": "<target-id>",
"snapshots": {
"<source url>": {"fetched_at": "<ISO time>", "summary": "<short>",
"key_facts": {"<fact>": "<value>"}}},
"reported": {
"<fingerprint>": {"id": "<finding id>", "first_seen": "<YYYY-MM-DD>"}}}
Keep summary short — this file is your memory, not an archive. Drop
reported entries older than 30 days; the app only sees about two weeks
of notes anyway.
Reply with a short digest. Your final reply is delivered to the task's
channel — usually the user's inbox — after every run, so it must be
readable at a glance. Plain text in the user's language, no preamble, no
JSON, no tables, no sign-off. Exactly one of these shapes:
- significant findings, one line each, at most five:
<Name>: <finding title> — <source url>
- only minor findings:
<Name>: <n> kleinere Änderungen, Details in der App.
- nothing new:
<Name>: nichts Neues.
- baseline run:
<Name>: Beobachtung läuft, Ausgangsstand gespeichert.
- the block could not be written:
<Name>: Lauf fehlgeschlagen — <reason in a few words>.
If you find nothing, write the block anyway, with "findings": []. The app
tells "nothing happened" apart from "the run did not take place".
C. The result block
The app reads only this block. Append it with memory
(action: "append", target: "daily") to the end of today's daily note.
The shape — a template, not data. Every <…> is replaced with real values;
the block you write must be valid JSON with no < or > left in it:
```watch
{
"targets": [
{"id": "<target-id>", "name": "<display name>", "kind": "<company|person>",
"focus": "<what is watched>", "schedule": "<human-readable, local time>",
"url": "<website>"}
],
"findings": [
{"id": "<finding id>", "target": "<target-id>",
"date": "<YYYY-MM-DD>", "severity": "<significant|minor>", "category": "<category>",
"title": "<one line, user's language>",
"detail": "<at most two sentences>",
"source": "<page where you saw it>"}
]
}
```
Write only what is real. targets comes from watch/targets.json;
findings only from changes you saw in this run (or []). A block that
repeats this template's placeholders, or invents a company that is not on
the watchlist, puts false alerts on the user's phone.
Rules that are not negotiable:
targets always holds the complete list from watch/targets.json —
not just this run's target — without the internal task_id. The app takes
the list from the newest note; an incomplete list makes targets disappear.
kind is company or person. schedule is a human-readable
label in the user's language and local time ("täglich 07:30"), not the cron
expression.
- A finding's
id is assigned once and never changes. Scheme:
<target>-<first-seen date>-<running number>, recorded in the state's
reported map under the finding's fingerprint. Never derive an id from
today's date for a change that was already reported — the app remembers
announced ids and would notify twice.
- Valid JSON: double quotes, no comments, no trailing commas. The app
discards a broken block — the whole run is then lost.
detail stays within two sentences. Daily notes are truncated at
20,000 characters when synced and the space is shared; long explanations
belong in watch/<id>.json, not here.
source is the page where you saw it, not the homepage. No verifiable
source, no finding.
What you do not do
- Do not invent findings to make a run look successful. An empty run is a
good result.
- Do not estimate prices or figures. If it is not stated, it is not in there.
- Do not report anything you only know from prior knowledge — only what you
actually saw in this run.
- Do not write anything beyond the digest in step 8. The report is the
block; the digest only points at it.
Prerequisites
- Cloud memory must be enabled for the agent instance. It is off by
default; without it the daily notes never leave the sandbox and the app
shows nothing. If a user asks why the app stays empty, point them to the
HybridClaw settings in the web workspace.
- A delivery channel for scheduled tasks (see step A.3).
1---2name: competitor-monitoring3description: Keep watch on competitors (and, where sources allow, industry people) on a schedule and report only real changes. Use when someone wants a competitor monitored, asks what a watched company has been doing, wants to change or stop a watch — and always when a scheduled run tells you to. Triggers include "watch", "keep an eye on", "monitor competitor", "what is the competition doing", "what changed at X", "stop watching", and their German equivalents ("beobachte", "behalte im Auge", "was macht der Wettbewerb").4user-invocable: true5metadata:6 source: https://skillmd.com/skills/hybridaione/competitor-monitoring7 hybridclaw:8 short_description: Scheduled competitor monitoring.9 category: business10 tags:11 - sales12 - competitors13 - monitoring14 - cron15 - memory16 related_skills:17 - search.news18 - search.web19---2021# Competitor Monitoring2223You watch competitors for a salesperson. They read the result in an app, not24in the chat — the Sales Companion iOS app parses a fenced `watch` block from25your daily memory note. Your job is not to write a nice report but to **detect26reliably what changed since the last run**, and to leave out everything else.2728Keep the tone factual. No "exciting", no "interestingly". A salesperson reads29this in twenty seconds in the morning. Write names, titles and details in the30user's language.3132## The four stores3334| What | Where | How |35|---|---|---|36| Schedule | one cron task per target | `cron` `add` / `update` / `remove` — chat only |37| Watchlist | `watch/targets.json` in the workspace | `read` / `write` |38| Comparison state | `watch/<target-id>.json` in the workspace | `read` / `write` |39| Result for the app | today's daily note | `memory` with `action: "append"`, `target: "daily"` |4041**`watch/targets.json` is the watchlist.** Scheduled runs cannot call `cron`42at all — the tool is blocked there — so they can only know the full list from43this file. Change it in the same step as the cron task, every time:4445```46{"targets": [47 {"id": "<target-id>", "name": "<display name>", "kind": "company",48 "focus": "<what to watch>", "schedule": "<e.g. täglich 07:30>",49 "url": "<website>", "task_id": <number from cron add>}50]}51```5253This and every other `<…>` shape in this file is a **template**: fill in54real values from the user's request, `watch/targets.json` or this run. Never55write a template, or an example company name from this file, into any file.5657`task_id` is the number `cron add` reports ("Scheduled … task #42 …"); pass58it as `taskId` to `update` and `remove`.5960Why these stores: the `memory` tool may only write today's daily note, and61only memory files are synced to the platform, where the app reads them. A62workspace file never reaches the app, so it holds your own state.6364## A. Set up a watch6566This happens in a chat with the user, where `cron` is available.67681. Clarify **who** and **what to look for**. If the focus is missing, ask69 exactly once ("What should I watch at Allianz — prices, products,70 people?"). If the website is missing, look it up yourself; do not ask.712. Derive a **stable id** from the name: lowercase ASCII, hyphens, no72 umlauts — `allianz-de`, `barmenia`, `jane-doe`. If `watch/targets.json`73 already has it, update that watch instead of adding a second one.743. Create the cron task:7576 ```77 cron action=add78 cron="<minute> <hour> * * *"79 tz="<user's IANA time zone>"80 channel="<the user's email address>"81 prompt="Competitor monitoring: use the competitor-monitoring skill,82 read its SKILL.md and follow it. Target: <target-id>83 (<display name>, <website>), kind=<company|person>,84 focus: <what to watch>."85 ```8687 - Pass `tz` with the user's IANA time zone so the run keeps its local time88 across daylight-saving changes. If you do not know it, ask; for a89 German-speaking user, `Europe/Berlin` is the usual answer.90 - `channel` is **required** when the request comes from a web chat91 session: the tool refuses to schedule there, because the output of a task92 created without a delivery channel would be discarded. Use the user's93 email address or another configured messaging target. If none is known,94 say so plainly instead of trying without one.954. Add the target with its `task_id` to `watch/targets.json` (create the file96 if it is missing).975. **Do not create `watch/<id>.json`.** The first scheduled run records the98 baseline; a file created now would only be empty.996. **Write the result block right away** (section C) with the updated list and100 `"findings": []`. Otherwise the new target only appears in the app after101 the first scheduled run.1027. Confirm in one sentence: "I'm watching Allianz from tomorrow morning,103 focus pet insurance."104105**Changing a watch** (time, focus): `cron action=update taskId=<task_id>`106from `watch/targets.json`, then update the entry there. Never add a second107task for the same target.108109**Removing a watch:** `cron action=remove taskId=<task_id>`, remove the110entry from `watch/targets.json`, delete `watch/<id>.json`, write the block111again.112113## B. A scheduled run114115You receive the task's prompt. `cron` is not available here; do not try it.116Work strictly in this order:1171181. **Read the watchlist and the previous state.** `read watch/targets.json`119 and `read watch/<id>.json`. If the state file is missing or its120 `snapshots` are empty, this is a **baseline run**: fetch the sources, save121 them as the state (step 6) and report **nothing** as a finding — there is122 nothing to compare against yet. Still write the block with123 `"findings": []`.1242. **Fetch the current state.** Use several sources, not just one:125 - `web_search` with `freshness: "week"` on company name plus focus126 - `web_fetch` on the pages that belong to the focus (pricing, product,127 press pages) — they are listed in the state under `snapshots`128 - for `kind=person`, also search name plus role. Social activity is only129 partly visible with the available tools, so rely on interviews, talks,130 press releases and job changes.1313. **Compare.** For each source: did the substance change? Ignore counters,132 dates, cookie banners, ordering and rewording that does not change meaning.1334. **Judge.** A finding only counts if it falls into one of these categories.134 Use the category names as given; the app shows them as labels.135136 | `category` | Example |137 |---|---|138 | `Preis` (price) | plan more expensive, new discount, tiers changed |139 | `Produkt` (product) | new product, benefit dropped, terms changed |140 | `Partnerschaft` (partnership) | cooperation, reseller, integration |141 | `Finanzierung` (funding) | funding round, acquisition, sale |142 | `Personal` (people) | change in management or sales leadership |143 | `Stellen` (hiring) | notable job ads that reveal a direction |144145 Everything else — guide articles, blog posts, social chatter, redesigns,146 anniversaries — is `severity: "minor"` or does not belong in the block at147 all. `severity: "significant"` is only for what the salesperson **must know148 today because it changes a conversation**; only `significant` triggers a149 notification on their phone. When in doubt, `minor`: a phone that buzzes150 too often gets muted.1515. **Deduplicate.** Give every finding a **fingerprint**: category, source URL152 and the new value in a few words — `Preis|https://allianz.de/preise|Premium153 27,90 €`. If the state's `reported` map already has that fingerprint, the154 change was reported before: drop it from this run. Otherwise assign a new155 id (see section C).1566. **Write the result block first** (section C). Only if the `memory` append157 succeeded, go on to step 7. If it fails — for example because today's note158 is full — stop and leave the state untouched, so the next run finds the159 same changes again instead of losing them. Reply with the failure line160 (step 8).1617. **Then update the state.** `write watch/<id>.json`:162163 ```164 {"target": "<target-id>",165 "snapshots": {166 "<source url>": {"fetched_at": "<ISO time>", "summary": "<short>",167 "key_facts": {"<fact>": "<value>"}}},168 "reported": {169 "<fingerprint>": {"id": "<finding id>", "first_seen": "<YYYY-MM-DD>"}}}170 ```171172 Keep `summary` short — this file is your memory, not an archive. Drop173 `reported` entries older than 30 days; the app only sees about two weeks174 of notes anyway.1751768. **Reply with a short digest.** Your final reply is delivered to the task's177 channel — usually the user's inbox — after **every** run, so it must be178 readable at a glance. Plain text in the user's language, no preamble, no179 JSON, no tables, no sign-off. Exactly one of these shapes:180181 - significant findings, one line each, at most five:182 `<Name>: <finding title> — <source url>`183 - only minor findings: `<Name>: <n> kleinere Änderungen, Details in der App.`184 - nothing new: `<Name>: nichts Neues.`185 - baseline run: `<Name>: Beobachtung läuft, Ausgangsstand gespeichert.`186 - the block could not be written: `<Name>: Lauf fehlgeschlagen — <reason in a few words>.`187188If you find nothing, write the block anyway, with `"findings": []`. The app189tells "nothing happened" apart from "the run did not take place".190191## C. The result block192193The app reads **only** this block. Append it with `memory`194(`action: "append"`, `target: "daily"`) to the **end** of today's daily note.195196The shape — a template, not data. Every `<…>` is replaced with real values;197the block you write must be valid JSON with no `<` or `>` left in it:198199````markdown200```watch201{202 "targets": [203 {"id": "<target-id>", "name": "<display name>", "kind": "<company|person>",204 "focus": "<what is watched>", "schedule": "<human-readable, local time>",205 "url": "<website>"}206 ],207 "findings": [208 {"id": "<finding id>", "target": "<target-id>",209 "date": "<YYYY-MM-DD>", "severity": "<significant|minor>", "category": "<category>",210 "title": "<one line, user's language>",211 "detail": "<at most two sentences>",212 "source": "<page where you saw it>"}213 ]214}215```216````217218**Write only what is real.** `targets` comes from `watch/targets.json`;219`findings` only from changes you saw in *this* run (or `[]`). A block that220repeats this template's placeholders, or invents a company that is not on221the watchlist, puts false alerts on the user's phone.222223Rules that are not negotiable:224225- **`targets` always holds the complete list** from `watch/targets.json` —226 not just this run's target — without the internal `task_id`. The app takes227 the list from the newest note; an incomplete list makes targets disappear.228- **`kind`** is `company` or `person`. **`schedule`** is a human-readable229 label in the user's language and local time ("täglich 07:30"), not the cron230 expression.231- **A finding's `id` is assigned once and never changes.** Scheme:232 `<target>-<first-seen date>-<running number>`, recorded in the state's233 `reported` map under the finding's fingerprint. Never derive an id from234 today's date for a change that was already reported — the app remembers235 announced ids and would notify twice.236- **Valid JSON**: double quotes, no comments, no trailing commas. The app237 discards a broken block — the whole run is then lost.238- **`detail` stays within two sentences.** Daily notes are truncated at239 20,000 characters when synced and the space is shared; long explanations240 belong in `watch/<id>.json`, not here.241- **`source` is the page where you saw it**, not the homepage. No verifiable242 source, no finding.243244## What you do not do245246- Do not invent findings to make a run look successful. An empty run is a247 good result.248- Do not estimate prices or figures. If it is not stated, it is not in there.249- Do not report anything you only know from prior knowledge — only what you250 actually saw in this run.251- Do not write anything beyond the digest in step 8. The report is the252 block; the digest only points at it.253254## Prerequisites255256- **Cloud memory must be enabled** for the agent instance. It is off by257 default; without it the daily notes never leave the sandbox and the app258 shows nothing. If a user asks why the app stays empty, point them to the259 HybridClaw settings in the web workspace.260- A delivery channel for scheduled tasks (see step A.3).